The implied long-term earnings growth of the S&P 500 is 5.74%. The return of the market is a function of dividends, earnings, and expectations (quantified by the P/E multiple). The P/E ratio looks extended and is a concern, and while cyclically adjusted market implied growth rate (CAIGR) is low versus its history, long-term earnings growth has also slowed.
Figure 1: S&P 500 Operating and Reported EPS
Figure 2: S&P 500 EPS and Price Return
Source: Spellman, FactSet, Standard & Poor’s
Since 1989, S&P 500 earnings per share have grown at about 5% (figure 1), with operating earnings growing slightly faster than reported earnings (5.3% versus 4.9%). S&P 500 EPS growth is about the same rate of nominal GDP. Operating earnings, which remove the impact of one time charges (mistakes of corporate management), are less volatile than reported earnings (standard deviation of 21% versus 120%), and recently the gap between the two numbers has widened. This is normal during earnings recessions when management writes off assets and tries to hide the real results. Earnings growth is tied to long-term price appreciation (figure 2); thus, a closer look at the market’s expectations for growth is warranted.
Figures 3-5: Drivers of S&P 500 Returns
Source: Spellman, FactSet, Standard & Poor’s
While long-term price appreciation is driven mostly by earnings, since 1997 yearly price changes have been driven by P/E more than EPS growth (figures 3-5). The R-squared between next 12-month P/E changes (using operating earnings) and S&P 500 price return is 0.70, versus only 0.29 for forward earnings growth.
Can the P/E continue higher and propel markets? The trailing 12 month P/E using reported earnings was 24.2 as of the end of September (figure 6). Excluding the internet bubble and recessions, this is the highest level since 1950. What makes this even more alarming is that earnings in the P/E ratio is elevated (figure 1).
Figure 6: S&P 500 TTM P/E (Reported)
Source: Spellman, Shiller (data only)
Shiller’s CAPE, or cyclically adjusted P/E, uses normalized earnings (a 10-year average). This is an improved P/E since price is a function of the normalized earnings power, investors look through temporary high and low earnings, growth, and risk. Ten years is a little long, but the purpose of the modification is to determine the normal earnings power over an economic cycle (about six years). The CAPE is also trading at a high (figure 7), and this does not bode well for future returns. Since 1987, the CAPE has had an inverse relationship with five-year forward annualized S&P 500 price returns (figure 8).
Figures 7 and 8: CAPE (left) and CAPE versus Returns (right)
Source: Spellman, Shiller (data only)
While the CAPE adjusts P/E for normalized earnings, it does not adjust for other variables that directly impact P/E. The Gordon Growth Model shows that price is a function of earnings, payout, risk, and growth. Thus, the price, and therefore P/E, is influenced by more than earnings.
P0 = E1 * payout ratio / (r – g)
P0 / E1 = payout ratio / (r – g)
The discount rate (“r” above) that we apply to earnings is heavily influenced by current interest rates. It is also influenced by inflation which impacts rates (since 1950, the 10-Year Treasury bond has averaged 1.87% above inflation). Figures 9 and 10 show the inverse relationship between CAPE and the 10-Year Treasury bond and CPI. Lower interest rates and inflation tend to drive up P/E, as theory would predict. However, moderate rates lead to the highest P/Es. Extremely low inflation and rates may be a sign of stress, which boost the equity risk premium (r = risk-free rate + equity risk premium) and negates the benefit of lower rates.
Figures 9 and 10: CAPE versus 10-Year Treasury Bond (left) and Inflation (right)
Source: Spellman, Shiller (data only)
The highest P/Es have occurred when interest rates were higher than today (about 2.48%), so higher rates that many investors expect to materialize may not negatively impact the P/E. Rates are also low relative to inflation (1.85%), which reflects the bubble in bonds induced by quantitative easing. While rates and inflation justify the current CAPE, be wary since the only time CAPE was higher since 1950 was during the internet bubble of the late 1990s.
The CAPE, while cyclically adjusted for earnings, does not adjust for the discount rate which can have a huge impact on P/E. To more fully adjust for the impacts of the economic cycle, let me introduce a new model called the cyclically adjusted implied growth rate (CAIGR). The CAIGR normalizes earnings, payout, and rates. CAIGR uses 10-year average earnings, 10-year Treasury bond rates (or inflation) plus an appropriate normal risk premium, and the 10-year average payout ratio. Using the Gordon Growth Model, one can solve for implied long-term growth.
P0 = E1 * payout ratio / (r – g)
r – g = E1 * payout ratio / P0
g = r – E1 * payout ratio / P0
CAIGR is growth where E1 = 10-year average EPS, payout ratio = 10-year average payout, and r = 10-year inflation rate plus a constant normal equity risk premium
The higher the implied growth rate, the higher expectations, and the more pricey the market.
The required rate of return needs further explanation.
Traditionally, the required rate of return (“r”) equals the 10-year Treasury bond rate plus an equity risk premium since stocks are more volatile than bonds; however, the 10-year Treasury bond rate is artificially low due to manipulations by the Federal Reserve’s quantitative easing program. Therefore, calculating a risk premium to inflation is preferable. Currently, the 10-year Treasury bond is 0.63% above the rate of inflation, compared to an average of 2.03% since 1950 and a median premium of 2.21% (figure 11). Thus, Treasury bonds appear to be overvalued. (Note: because of this, if we utilized the T-bond in the CAIGR instead of inflation to compute the required return, then the implied growth rate calculated would appear to be low.)
Figure 11: 10-Year Treasury Bond versus CPI Y-Y % Change
Source: Spellman, Shiller (data only)
S&P 500 stocks outperformed 10-year Treasury bonds by 4.62%, 3.42%, and 2.30% from 1928-2016, 1967-2016, and 2007-2016, respectively (Damodaran, New York University, www.damodaran.com). If we assume that stocks outperform bonds by 4%, and bonds yield 2% more than inflation, then the appropriate equity risk premium to add to inflation is 6%. Since the inflation rate varies from year to year and the market should look through temporary imbalances, 6% is added to a 10-year average of the inflation rate to calculate the required rate of return.
Figures 12 and 13: CAIGR versus EPS (as Reported) Growth using 10-year Averages (left) and 5-year Averages (right)
Source: Spellman, Shiller (data only)
CAIGR using the 10-year inflation methodology is show in figure 12. CAIGR using five year averages (for inflation, EPS, and payout) – five years is approximately the length of one economic cycle – is illustrated in figure 13. The current – using the 10-year methodology – CAIGR reading is 5.74%. Given low rates of productivity, low population growth, and moderate to low inflation, the current CAIGR reading of 5.74% is realistic to optimistic despite the fact that it is lower than historical averages. Still, this presents a slightly better picture of future returns and market expectations than the CAPE (figures 7 and 8) since the CAIGR normalizes all components of price (earnings, rates, risk premium, and payout) and “r” is also low, whereas the CAPE simply normalizes earnings.
The current CAIGR of 5.74% is above the historical actual rate of EPS growth (about 5%). This is not atypical as the CAIGR has been overly optimistic since the 1960s. It has gyrated somewhat in tandem with average of actual 10-year EPS growth (figures 12 and 13). The highest implied growth occurred in the early 1980s when inflation was elevated (higher inflation drives up the required rate of return and boosts calculated implied growth); however, this was also the period with the highest EPS growth (high inflation yields high nominal growth rates) of the last 66 years. Over the last 20 years, the highest CAIGR (about 8%) was during the internet bubble, the second highest was during the financial bubble (about 7%), and third highest was in mid-2014 (about 6.3%) at the same time as 10-year average of actual growth drove higher (to around 6.1% versus 4.7% right now).
Unlike the CAPE, there is low explanatory power between the level of the CAIGR and future returns. On the other hand, change in CAIGR is much better than change in CAPE for forecasting future returns (figure 14). Five-year change in CAIGR has been negatively related to five-year forward annualized S&P 500 price returns since 1987. CAIGR has dropped from 6.3% in mid-2014 to 5.74% now; however, the five-year change in CAIGR is only -0.2%. Thus, this suggests quite modest returns.
Figure 14: Change in CAIGR and Future Returns
Source: Spellman, Shiller (data only)