1. The extant finance and economic literature lacks a commonly acceptable theoretical model to explain the equilibrium level of market valuation ratios, for which the equilibrium level would be the level to which the CAPE ratio should migrate. Therefore, in the context of CAPE, we are not using “equilibrium” in the strict finance definition of the word. Rather, we are observing an empirical normal level, toward which the CAPE ratio appears to mean revert. We also assume throughout the article that this equilibrium level may be time varying and influenced by a number of economic factors.
2. We match the log of the CAPE to the subsequent S&P 500 return. This is not an accident or data mining; the same annualized price change—over and above growth in earnings and dividends—is needed in order to move the CAPE market valuation ratio from 5x to 10x as from 25x to 50x. Therefore, the log of the CAPE is the correct predictor for future long-term annualized market returns.
3. Boudoukh, Richardson, and Whitelaw (2008) were the first to demonstrate the market is actually quite predictable in the short run, with the slope of short-term predictability higher than for the longer term; however, significant uncertainty overwhelms the predictive power. The fact that the slope is steepest at the one-year horizon is often overlooked.
4. Our research has explored demographic drivers for shifting equilibria in dividend yields, hence, indirectly in CAPE. Higher numbers of mature working-age adults (ages 40–60) go hand in hand with higher equity valuation levels and lower yields. Higher numbers of senior citizens correlate with sharply lower valuation levels and higher yields. Our research suggests a crossover, with a doubling of yields, in the coming 15 years.
5. Our colleagues Aked, Mazzoleni, and Shakernia (2017) showed that economic volatility is powerfully linked to the CAPE ratio. When economic volatility is low, the natural CAPE ratio can be much higher. The Achilles’ heel in this relationship is that low economic volatility is unlikely to remain low indefinitely. Economic volatility exhibits strong mean reversion.
6. Recall American economist Irving Fisher’s famous observation on October 17, 1929, that “stock prices have reached what looks like a permanently high plateau.” He uttered these words three months after the market peak, just four days before the Crash of 1929 got under way and the market plummeted almost 30% in a single week. Although we cannot confidently declare a “permanently high plateau,” we can assert with some confidence that if high valuations are sustained, then lower long-term future returns are likely to be the new normal. GMO’s Inker (2014) differentiates between a “purgatory,” in which markets plunge to valuation levels that permit respectable subsequent returns, and a “hell,” in which valuations remain lofty and future returns are permanently impaired.
7. In real terms, however, earnings are still below their 2014 peak; we’ve merely recovered part of the 2014–2016 earnings slump.
8. This argument can play out two ways. The point of using the 10-year average in the CAPE ratio is to smooth out earnings peaks and troughs, so the PE ratio is not distorted by either current peak or current trough earnings and creating an aberrant understated or overstated, respectively, PE ratio. In this sense, Siegel’s argument seems somewhat at odds with the core purpose of the CAPE. A few counters to his argument bear mention. Most 10-year spans encompass two earnings downturns so that one unusually deep downturn does not necessarily distort the CAPE. If, in the next two years, we do not have an earnings downturn, then the CAPE ratio will consist only of strong earnings with no recessionary lows, for the first time ever, leading to an artificially depressed CAPE. Finally, if we simply replace the average of 10-year real earnings, the CAPE dividend, with the median of 10-year real earnings, and thus neutralize the effect of artificially low earnings, the CAPE series does not change much from the standard series.
9. Real per share earnings were 40% higher in 1940 and 120% higher in 1950 compared to 1945.
10. Tower (2011, 2013) has examined many of Siegel’s proposed corrections to the CAPE ratio. He created alternative histories for CAPE, incorporating the proposed corrections in the historical data, so that the adjusted current CAPE could be compared with a similarly adjusted historical CAPE. He found that with this “corrected” apples-to-apples comparison, the predicted long-term future return from the corrected CAPE measures differed from the forecast of the standard CAPE measure by less than two percentage points, in all cases.
11. The situation is even more striking in Japan. A remarkably little-known fact is that Japan’s GDP per working-age adult has been growing faster than that of the United States or Western Europe, even though Japan’s GDP growth seems stalled. How can this be? Sluggish GDP growth, in a context of a shrinking working-age population, is actually rather impressive!
12. Many in the economics profession fail to grasp this basic truism: The growth of all existing businesses cannot match GDP growth because new enterprises will dilute their role in the future private sector. Extant businesses will not compose 100% of the future private sector. How fast does this happen? Of the 20 largest market-cap companies in the United States, five, composing 36% of the market cap of the top 20, did not exist as publicly traded businesses 30 years ago. Two, composing 12% of the market cap of the top 20, did not exist as publicly traded businesses 15 years ago. This would suggest that existing businesses have seen their share of aggregate market value diluted to 64% of their starting share in just 30 years as a consequence of entrepreneurial capitalism—a 1.5% dilution of existing businesses’ share of market-cap per year (See Bernstein and Arnott, 2003).
13. One of our favorite observations is that in 1820 a message could go from one city to the next at the speed of a horse, the same speed that had prevailed for millennia. But in only a decade, between the invention of the commercially viable railway in the 1820s and the telegraph in the 1830s, a message could be transmitted at the speed of light. Talk about disruptive technologies!! Railroads, telegraph and electric companies, radio and automobile companies were the internet fliers of their eras.
14. Of course, a comparison of incomes between 1980 and today can be quite challenging. The CPI has a hedonic adjustment to reflect this type of challenge, but real wages, despite such adjustment, may not fully capture rising life expectancies or improved quality of life. The downside is that many innovations crush the wages of unskilled labor and can show up as a drop in GDP. Perhaps more important is that the lack of growth in median income emphasizes the rising inequality in wealth distribution, which can put strong political pressure on profits as a means to satisfy calls for a more-equal distribution of wealth.
15. CAPE also shows promise in emerging markets, but over much shorter time spans, so we are ignoring this region in our analysis.
16. This is based on our publicly available equity valuation methodology, which compares CAPE to a target value half way between the current value and the long-term average. More detail is available at https://www.researchaffiliates.com/content/dam/ra/documents/asset-allocation/aa-methodology-822.pdf.
17. Brightman, Masturzo, and Beck (2015) offer a long-term comparison of valuation metrics for the US market.
18. A number of structural reasons—for example, different accounting conventions—can explain why a particular valuation ratio indicates different relative valuation levels from one market to another. That said, if multiple models, each with its unique idiosyncratic bias, all point in the same direction, the confidence that the US market is expensive or cheap relative to other markets strengthens the conclusion, and suggests the conclusion is not a result of mismeasurement specific to CAPE.
19. Actuaries, however, are still required to forecast future stock and bond market returns based on extrapolating past returns.
20. For skeptics of the early data, the correlation in the 70 years since World War II is 60%, while the correlations for EAFE and for the emerging markets, over the past 35 and 25 years, respectively, are north of 75%.