Over 70 years ago, Benjamin Graham and David Dodd proposed valuing securities with earnings smoothed across multiple years. Robert Shiller popularized this method with his version of the cyclically adjusted price-to-earnings ratio (CAPE) in the late 1990s, and issued a timely warning of poor stock returns to follow in the coming years. We apply this valuation metric across approximately 40 foreign markets and find it both practical and useful. Indeed, we witness even more examples of bubbles and busts abroad than in the United States. We then create a trading system to build global stock portfolios based on valuation, and find significant outperformance by selecting markets based on relative and absolute valuation.
The Futility Of Forecasting
Investors spend an inordinate amount of time and effort forecasting stock market direction, often with very little success. The conventional efficient market theory is that markets are not predictable and cannot be forecasted. Value has no place in the efficient market ivory tower, but does it seem reasonable for an investor, or perhaps a retiree, to have allocated the same amount of a portfolio to stocks in December 1999 that they did in 1982? Of course not.
However, valuation is best used as a strategic guide rather than as a short-term timing tool. It is most useful on a time scale of years and decades rather than weeks and months (or even days). While we can formulate a hypothesis for where the S&P 500 “should” be trading, the animal spirits contained in the marketplace invariably cause prices to deviate quite substantially from “reasonable” levels, often for years and even decades.
There are numerous models to consider when valuing stock markets, and a great summary can be found in a publication by The Leuthold Group titled “Stock Market Valuation: What Works and What Doesn’t?” The paper covers a number of models, including price-to-earnings (P/E) on trailing 12-month earnings per share (EPS), P/E on five-year normalized EPS, return on equity (ROE)-based normalized EPS, dividend yield, price-to-book, price-to-cash flow and price-to-sales. In general, they find that many of these metrics are decent at forecasting stock returns. Other models include the Q-ratio, and market capitalization to GNP/GDP (Buffett’s favorite). Another great summary is set forth in the paper “Estimating Future Stock Market Returns” by Adam Butler and Mike Philbrick.
However, we are not going to summarize all of the stock valuation models in existence; rather, we will focus on just one.
A Simple Model: 10-Year Normalized Earnings
Benjamin Graham and David Dodd are universally seen as the fathers of valuation and security analysis. In their 1934 book “Security Analysis,” they were early pioneers in comparing stock prices with earnings smoothed across multiple years, preferably five to 10 years. Using backward-looking earnings allows the analyst to smooth out the business and economic cycle, as well as price fluctuations. This long-term perspective dampens the effects of expansions as well as recessions.
Robert Shiller, the author and Yale professor, popularized Graham and Dodd’s methods with his version of this cyclically adjusted price-to-earnings ratio (CAPE).1
One common criticism of CAPE is that the measurement period of 10 years is too long. Critics claim recessions and expansions have an outsized impact long after they have faded from memory.2 Those critics also claim adjustments to CPI and accounting rules render comparisons across decades, or even centuries, meaningless. We agree there may be some variation, but later in the paper we examine CAPE in approximately 40 foreign markets with supporting results with regard to consistency.
Figure 1 is a chart of CAPE going back to 1881. The long-term series spends about half of the time with values ranging between 10 and 20, with an average and median value of about 16. The all-time low reading was 5, reached at the end of 1920, and the high value of 45 was reached at—you guessed it!—the end of 1999.
Asset allocators that believe in efficient markets allocate the same percentage of assets to equities when valuations are high as they do when valuations are low. But does that seem even remotely reasonable looking at Figure 1?
The 10 Best, And Worst, Times In History To Invest
To illustrate this point, we examined all year-end periods with a holding period for the next 10 years. What have been the 10 best, and worst, years to invest since 1871? Figure 2 details these years and their corresponding 10-year compounded real returns.
Many of the best starting points seem obvious in retrospect. 1948 and 1949 were great entries, preceding the Nifty Fifty mania, and of course 1918-1920—right before the Roaring Twenties—are on the list. 1988 and 1989 certainly would not be left out, with the Internet bull market ahead, as well.
The same hindsight applies for the bad years, as they often fell at the end of these massive bull runs. Bear markets set the stage for future bull markets and vice versa.
One simple takeaway from Figure 2 is the valuations at the start of these 10-year periods. The average valuation for the 10 best years was 10.92. The average valuation for the 10 worst years was 23.31, double that of the best starting points.
Buy Low, Sell High
Figure 3 is a table of all of CAPE year-end readings from 1881-2011. We list how often they occur, as well as the real forward returns. The red bar in Figure 4 is where we find ourselves as of the summer of 2012.
What we find is no surprise: It very much matters what price one pays for an investment! Indeed, it is an almost perfect stair step: Future returns are lower when valuations are high, and future returns are higher when valuations are low.
While more sophisticated models can be built, Figure 5 simply shows the shockingly similar trend lines of an inverse CAPE and future 10-year real stock returns.3
Valuation And Inflation
Besides general sentiment, what might cause this large variation in what multiples investors are willing to pay for stocks? After all, at a current value of around 1374, this means the S&P 500 could trade at either 315 or 2800 based on historical low and high multiples of 5 and 45, respectively. It is difficult for most investors to comprehend the possibility of stocks declining 80 percent or increasing over 100 percent, but both of these multiples have occurred in the past.
One of the determinants of the valuation multiple investors are willing to pay is the inflation rate as seen in Figure 6. The red bar is where we find ourselves as of the summer of 2012. When inflation is in the 1-4 percent “comfort zone,” investors are willing to pay a valuation premium compared with when there is either high inflation or outright deflation.4
There is very little in the literature regarding global CAPEs for international equity markets.5 We examined 39 countries with data from MSCI and Global Financial Data, including as much data as we could find, although there is some bias in the study. All the returns are real dollar returns.
Two countries had a century’s worth of data (U.S. and the U.K.), but most of the other countries go back to the 1970s and 1980s (see Figure 7).
We examined all the countries on a yearly basis since 1980, CAPE levels and future returns. The sample includes approximately 10 countries in 1980, 20 in 1990 and 30 by 2000. The results are in Figure 8 and largely confirm the U.S. data: Buy low, sell high.
We found most CAPEs averaged around 15-20, bottomed out around 7, and maxed out around 45 (a few made the U.S. bubble in the late 1990s look pathetic in comparison, as when Japan reached a value of nearly 100 in 1989).
A Global Stock Trading System
But can we turn this into a trading system? There is evidence that sorting countries on other measures of value works well. A good summary of the dividend literature can be found in the Tweedy, Browne paper titled “The High Dividend Return Advantage.”6 In the paper, they summarize a 1991 study by Michael Keppler titled “The Importance of Dividend Yields in Country Selection“7 that found that ranking the universe of countries by dividend yield also resulted in outperformance. He found that the highest-yielding countries outperformed the lowest-yielding from 1969-1989 by more than 12 percentage points per year.
Running a similar study using a different database (Global Financial Data),8 we sorted countries by quartiles from 1920-2011, beginning with nine countries and expanding to 18 by study end. We found that countries in the highest-dividend-paying quartile outperformed the countries in the lowest-paying quartile by 11 percentage points per year. (Also see the Appendix for tests on book value, dividends, cash flow and earnings.)
We then set out to test CAPE in a similar manner. Starting in 1980, we sort all countries by CAPE, and invest in the most undervalued x percent, rebalanced yearly. We also show the effects of investing in the most overvalued x percent, as well as a long/short portfolio. These returns are real returns net of inflation, and with yearly data (which will naturally understate drawdown figures). The sample includes approximately 10 countries in 1980, 20 in 1990 and 30 by 2000. Investing in the cheapest countries produces 2 to 7 percentage points of outperformance, while the overvalued countries underperform (see Figure 9). The spread is approximately similar to those appearing in the previously mentioned dividend studies, albeit slightly lower. However, investing in the cheapest countries on a relative basis does not protect the investor when all countries are expensive in a global equity bubble like 1999. We repeated the study, but only invested long if the country was below a CAPE of 15, and only short above a CAPE of 30. If the country does not qualify for the valuation filter, then that part of the portfolio sits in cash (although we do not receive any interest income in this test) (see Figure 10).
For the most part, adding the absolute CAPE-level filter results in better performance with lower drawdowns. This is to be expected, as the portfolio could be sitting in 20, 50 or even 100 percent cash (as with 1999 or 2007). In this case, the returns are higher as well. As many investors look at this table and salivate over the prospect of 15 percent real returns, recall Figure 7 and note that most of the cheapest countries fall in the troubled eurozone. How many investors have the stomach to invest in these countries with potential for the markets to get even cheaper? How many professional investors would be willing to bear the career risk associated with being potentially wrong in buying these markets?
Figure 11 depicts the equity curves from taking the cheapest 33 percent of countries (also with filter), the most expensive 33 percent of countries and the equal-weight benchmark.
Warren Buffett famously said, “Price is what you pay. Value is what you get.” Over periods of years and decades, it is evident that an investor’s real return is heavily dependent on the price paid for the asset. Investors can use CAPE valuation as a guidepost for both opportunities arising from negative geopolitical events as well as a sanity check against bubbling stock markets. Comparing global equity markets on a relative basis allows the portfolio manager to create portfolios of cheap stocks markets, while avoiding or even shorting expensive markets.
Appendix: Other Valuation Models
Samuel Lee has a great article titled “The Hedgehog’s Error”9 on Morningstar’s website that sorts global countries based on value (price/book) using the French/Fama database. Not surprisingly, he finds that sorting on value works well.
We utilize the database to sort the countries (12 in 1975 and rising to 20 by 1991) based on various measure of value. In Figure 12, we demonstrate the results of sorting the countries on a yearly basis and choosing the cheapest x percent of the universe (from 10 to 33 percent). Results are U.S. dollar based, nominal.
1 Shiller maintains a website with an Excel download that includes historical data with formulas illustrating how to construct his 10-year CAPE: http://www.econ.yale.edu/~shiller/data.htm. For a step-by-step guide, Wes Gray at Turnkey Analyst has a good post that walks through the steps necessary to construct the metric: http://turnkeyanalyst.com/2011/10/the-shiller-pe-ratio/
2 “Estimating Future Stock Market Returns” by Adam Butler and Mike Philbrick tackles the issue of different measurement periods from one to 30 years (as well as other valuation models).
3 John Hussman has a few good articles on this topic: “Estimating the Long-Term Returns on Stocks” and “The Likely Range of Market Returns in the Coming Decade”; Joachim Klement also recently published the paper “Does the Shiller-PE Work in Emerging Markets?” that performs a similar analysis.
4Rob Arnott of Research Affiliates touches on this important topic in his white paper “King of the Mountain” (http://www.researchaffiliates.com/Our%20Ideas/Insights/Fundamentals/Pages/F_2011_Sept_King_of_the_Mountain.aspx). Two other books speak of CAPEs and inflation/deflation levels. The first is “Unexpected Returns: Understanding Secular Stock Market Cycles” by Ed Easterling, and John Mauldin’s “Bull’s Eye Investing: Targeting Real Returns in a Smoke and Mirrors Market.”
5 One such resource is Russell Napier, who authored Anatomy of the Bear: Lessons From Wall Street’s Four Great Bottoms, and who discusses global CAPEs in a video here: http://video.ft.com/v/946244201001/Long-View-Historian-sees-S-P-fall-to-400 . We also found two great recently published papers: “Does the Shiller-PE Work in Emerging Markets?” by Joachim Klement (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2088140), and “Value Matters: Predictability of Stock Index Returns” by Angelini, Bormetti, Marmi and Nardini (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2031406).