Why Valuation Doesn’t Insure Against A Significant Market Decline
Given a clearly overbought market, the re-emergence of Europe’s sovereign debt problem and the Fed reducing the imminence of QE3, even the bulls concede that a correction is likely. Overall, however, investors remain optimistic, and are looking forward to any correction as a buying opportunity, maintaining that the economy is too strong and the market too cheap to decline very much. As we have written about in recent comments, we do not think the economy is anywhere as strong as many believe. Moreover, we do not accept the conventional wisdom that the market, at current levels is undervalued, a point we want to make in this comment.
Most strategists and economists measure value by multiplying forward-looking estimates of operating earnings by 15 and comparing the result to the current level of the index they are following, most often the S&P 500. Studies show, however, that this method has poor predictive results in estimating future stock market returns. We, on the other hand, use 15 times cyclically-adjusted trailing reported (GAAP) earnings to estimate the reasonable value the market, a method with far greater success in anticipating long-term returns.
Reported (GAAP) earnings are calculated in accordance with Generally Accepted Accounting Principles and are the official earnings required to be submitted to the SEC. On the other hand, almost all companies, when reporting earnings in their press releases, headline so-called "operating earnings" that are also used by most Wall Street analysts. These are also the earnings picked up by organizations that compile the widely-used consensus earnings forecasts. Operating earnings throw back into earnings a large number of expenses considered by management as non-recurring, including such items as severance pay, plant closings, plant start-ups, inventory write-downs and any number of other expenses that corporations may want to write off in order to make earnings look better. In the past 15 years companies have become a lot more creative about what items they can write off, and now a large number of expenses that used to be considered normal are called unusual even when these write-offs are taken year after year.
Another problem is applying a 15 multiple to operating earnings in order to estimate fair value of the S&P 500. The widely-used multiple of 15 times is based on long-term studies from the mid-1920s to the late 1990s showing that number to be the average multiple over a long period. But the earnings results used in those studies were trailing GAAP earnings, not forward-looking operating earnings, which did not come into vogue until the mid to late 1980s. Since operating earnings are always higher than reported earnings, the multiple used on operating earnings should be substantially lower than the multiple applied to reported earnings.
The problem is exacerbated by the fact that estimates of year-ahead operating earnings are almost always wrong, and most often on the high side. Since 1999 such forecasts have missed their target by about 23%. At year-end 2007 the consensus estimate of S&P 500 operating earnings for 2008 was $89, and remained there until the end of May. Even at the end of October, only two months from year-end, the estimate was at $72. The actual number came in at slightly under $50 just a short while later.
The estimate of operating earnings for 2009 was even more laughable. In May 2008, the estimate for 2009 was as high as $110. At year-end 2008, months after the credit crisis was already known, the number had only come down to $99. It ended up far lower at $57. Given all of the problems with using forward operating earnings rather than trailing GAAP earnings it is no wonder that studies indicate that, on average, the price-earnings ratio on forward operating earnings is only about two-thirds of the ratio on trailing reported earnings.
Another adjustment we make is the use of cyclically-adjusted (smoothed) reported earnings. Earnings over the course of a business cycle are highly volatile and constantly fluctuate from high to low and back again. Therefore, applying a P/E ratio to any specific year can be highly misleading. Corporate profit margins from 1952 to 2009 have fluctuated between slightly below 4% to slightly above 8%, but are now over 10%. Historically, margins have always reverted to the mean. Smoothing the trend of earnings over a period of time eliminates the distortions arising from extreme highs and lows.
The current 2012 consensus for S&P 500 estimated operating earnings is about $100. Multiplying this estimate by 15 results in estimated fair value of the index at 1500, leading most strategists to conclude that the market is undervalued. On the other hand, by using our estimate of cyclically-adjusted trailing earnings of $76, the market is overvalued by 18%. Moreover, past secular bear markets as well as many cyclical bear markets have historically bottomed at P/E ratios of 10 or under, a level that is far from unusual. In fact, the S&P 500 has sold at multiples of 10 or below in 17 different years since 1950. Therefore, if the "sweet spot" that the market has been enjoying is coming to an end, as we maintained in last week’s comment, valuation will not provide any cushion against a substantial decline in the period ahead.