Embedded Predictions and the Valuation Debate

savita chart




Traders returning to their desks after the long holiday weekend will be greeted by the continuing Greek saga, a guessing game about when the Federal Reserve will raise rates and a megamerger in the cable industry. But the debate I am much more interested in is the one taking place about U.S. stock valuations. It is more significant, and not as well understood as those other discussions.

The bears have been saying for some time now that stocks are at best fully valued, meaning there isn’t much room for gains after equities more than tripled from their March 2009 lows. The ursine argument is that low future returns are a given, and an ugly crash is the worst-case scenario.

The bulls argue that ultralow inflation allows for higher valuation metrics based on price-earnings ratios or Shiller’s CAPE (cyclically adjusted P/E). Compared with all historic medians, these measures are rather high. But compared with similar eras of very low rates and very low inflation, they are quite reasonable.

Meanwhile, the market keeps meandering along, occasionally reaching new highs.

I want to suggest another framework for evaluating the market’s prospects. It involves future earnings growth and a range of possible outcomes. Some may find this approach to be unsatisfactory, for it is dependent upon future events that are both unknown and unpredictable.

Have a look at the table above: It comes from Bank of America Merrill Lynch equity and quant strategist, Savita Subramanian. I have found that many investors focus on the implied 10-year annualized returns, rather than upon the range embodied by the 90 percent confidence interval.

Placing emphasis on specific expected returns embeds an assumed prediction about future earnings. But there are too many variables that affect future corporate profitability to make a specific guess with a high degree of confidence. Hence, many valuation arguments with an embedded forecast are usually binary in their outcome. The guess is either right or wrong; the correlated investment posture is either a winner or a loser.

Focusing on a range allows us to recognize that several factors remain unknown at present. These are crucial in determining future equity valuations.

Consider these variables:

— Will the U.S. economy slip into a recession? Begin to accelerate from its long slow recovery from the financial crisis? Or will it keep muddling along?

— New unemployment claims are at lows not seen in a generation. Will this create wage pressure?

— If wages rise, will consumers save or spend their pay increases?

— Will companies expand their modest capital expenditure budgets? Will research and development spending increase?

— Are corporate sales going to accelerate anytime soon?

— Will corporate profits be hurt by increased capital expenditure and wage increases? Or will a virtuous cycle raise profits?

Each of the above unknowns has the potential to affect U.S. equity valuations. This is why a range of possible outcomes is a better way to think about valuations than declaring stocks cheap, expensive or fairly valued.

Hence, when we consider valuation metrics, we should be aware that many of us build a forecast into those ratios. We don’t know what future earnings will be in 2016 or the year after, and merely extrapolating a number from present levels is at best guesswork.

To clarify, I am not suggesting stocks are especially cheap here or that equities must go up. Rather, I am emphasizing that since we have only limited ability to estimate future earnings, the reliance on valuation metrics such as P/E ratios should recognize the likely range of outcomes.

Stocks, like the Fed, are data-dependent. It is a shame we don’t hear that admission from more market participants.


Originally published at: 

A Better Way to Think About Stock Prices



Print Friendly, PDF & Email

What's been said:

Discussions found on the web:
  1. Iamthe50percent commented on May 26

    Please excuse my ignorance, but what does the abbreviation “RSQ” stand for?

    • Mike Radigan commented on May 26

      RSQ is R-Squared – the percentage of the variation in y that is explained by x. It is a measure of diversification that determines how closely performance parallels an appropriate market benchmark over a period. The market is understood to have a R-Squared of 100%. A R-Squared of say 95% contains 95% of the market’s diversification and risk.

    • Iamthe50percent commented on May 26

      Thank You. I know what R-squared is, just never saw the abbreviation in my math classes.

    • DonQuixote commented on May 26

      I also don’t know for what RSQ stands. The best answer I found is that the “RS” stands for R-Square that, for investing, measures the correlation between an index and a benchmark. I have no idea what the “Q” means. An explanation of RS appears at http://www.investopedia.com/terms/r/r-squared.asp

      My first thought on seeing this post was that “Life is uncertain – Eat dessert first.” This may not help you pick an investment, but it will hopefully adjust your attitude after a bad day in the market.

    • Alex commented on May 26

      My guess is it is the Coefficient of Determination, which tells you how well the regression model fits the actual data.

  2. DHM commented on May 26

    Mr. Ritholtz,

    I believe the table has an error. The median/average at the bottom is underneath the SE column. Shouldn’t it be underneath the Annualized Return column?

  3. scone commented on May 26

    If these numbers are nominal, then you’re still not making much, once you account for inflation and taxes. Say you get a low return, 6% or less, with inflation at 2% or more, and taxes removing another 1% or more. In that kind of scenario, you have to save hugely more to meet your goals, or cut down your goals, which suggests even more low growth. Vicious circle.

    • Terry commented on May 26

      …and so what are the “better” options?

  4. rd commented on May 26

    A question about the low inflation, high PE periods. How many of those data points coincide with recessions where stock prices are low but earnings dropped more, so the PE is high (2009?, 1932?) temporarily before earnings recover. How do those compare with record market highs at peak earnings margins?

    How are upcoming demographic shifts with boomers leaving the work force, drawing down their savings, while Generation X creates a hole in the critical peak productivity 35 to 55 demographic going to impact the markets? How will the stock market PE multiples react to this likely slower growth scenario for a decade or so before the larger millenial bulge moves into their 40s while the percentage of population over 65 starts to decline? The post WW II stock market has played out with a demographic back drop where the baby boom demographic bulge drove a lot of economic growth over the past 40 years. Extrapolating that to the next decade may be difficult – there are long-standing trend lines that go back a century but the economy, stock market, and interest rates tend to occupy zones above or below that trend line for one to decades at a time.

  5. neilmcgovern commented on May 26

    I might be reading this incorrectly, but the Shiller numbers are more like the PB/V numbers (27.0, 16.6) based on the link on the text.

  6. scone commented on May 26

    The markets don’t have to give you better options. You can go through long periods of “meh,” particularly after several bull years.

Posted Under