The chart is based on data through the end of 2012. Smithers notes “At that date the S&P 500 was at 1426 and US non-financials were overvalued by 44% according to q and quoted shares, including financials, were overvalued by 52% according to CAPE. With the S&P 500 at 1552 the overvaluation was 57% for non-financials and 65% for quoted shares.”
Unfortunately, that seems about right. Let’s translate this into an estimate of prospective 10-year total returns, assuming underlying nominal economic growth rate of about 6.3% (which may be optimistic, but is a robust peak-to-peak norm across economic cycles, and is unlikely to be pessimistic), and a dividend yield of about 2.2% on the S&P 500. With that, a 65% overvaluation in quoted shares, reverting to fair valuation a decade from now, would imply a 10-year annual nominal total return on the S&P 500 of 1.063*(1/1.65)^(1/10) + .022 – 1 = 3.3% annually. That’s right in line with the estimates we obtain from a wide range of other historically reliableapproaches (historically reliable in italics, because the “Fed Model” is not).
Notice that in 1982, the -0.7 reading on Smithers’ log-scale chart implied that stocks were undervalued by exp(-0.7)-1 = -50%. At that point, with the dividend yield on the S&P 500 about 6.7%, one would have estimated a 10-year prospective total return for the S&P 500 of 1.063*(1/0.5)^(1/10)+.067 – 1 = 20.6% annually. One would have been correct.
In contrast, note that in 2000, the 1.0 reading implied that stocks were overvalued by exp(1.0)-1 = 172%. At that point, with the dividend yield on the S&P 500 at just 1.2%, one would have estimated a 10-year prospective total return for the S&P 500 of 1.063*(1/2.72)^(1/10)+.012 = -2.6% annually. Again, one would have been correct.
With due respect to Howard Marks and Warren Buffett
At present, we estimate a 10-year total return on the S&P 500 over the coming decade averaging just 3.5% annually, with zero total returns over a horizon of about 7 years, and expected losses for the S&P 500, including dividends, over shorter horizons.
…..The last four years of market advance have reduced FUTURE retruns.
While our estimates for 10-year total returns exceeded 10% annually near the 2009 market lows, the recent advance has, in effect, “eaten” most of those prospective returns. The well-admired bond manager Howard Marks is very correct when he notes “appreciation at a rate in excess of the cash flow accelerates into the present some appreciation that otherwise might have happened in the future.”
Where I differ from even Howard Marks and Warren Buffet here, is that if you are going to rely on a summary measure in order to value long-lived assets like stocks (both Marks and Buffett point to “forward operating earnings” today), that summary measure must be representative of the long-term stream of cash that investors can expect to receive over time. The hook today is that investors are using analyst estimates of next year’s operating earnings as if they are representative of the entire long-term stream, and that this one number can be used as a “sufficient statistic” for long-term corporate profitability.