Maybe It's Not Different This Time

By WARREN JACOBS | MORE ARTICLES BY AUTHOR

Seen historically, and not just in the context of the last 20 years, valuation levels of the S&P 500 index may still be very elevated.

• Article
• Comments (16)

 

FINANCIAL PROFESSIONALS PROPOSING risky investments are fond of challenging anyone who disagrees with their promise that "this time it's different." While risk may be appropriate for those just beginning their investment journey, twilight investors approaching the day they must live off their savings should be especially interested in asking what will happen if this time it's not different.

Historically, investors have relied on price-to-earnings, price-to-dividends and price-to-book ratios to value the stock market. Based on their lofty levels in the past two decades, many financial professionals have chosen to ignore dividend and book-value ratios. And they have used estimates of future earnings, rather than actual past earnings, in constructing their P/E ratios. These ratios are new and different, even if the underlying accounting is not.

Relevant P/E ratios require reliable company profit levels to compare with the stock index price. Many financial professionals base their recommendations on projections of future earnings for the Standard & Poor's 500 stock index.

I examined the reliability of analysts' 12-month forward S&P 500 earnings projections by comparing Thomson Financial's analyst estimates with the actual S&P profits from Robert Shiller's Website (http://www.econ.yale.edu/~shiller/). Estimates of the next 12 months' earnings from January 1979 through October 2003 ranged from 23% below to 59% above the actual S&P earnings. The 298 Thomson monthly analysts' projections averaged 19% greater than the actual S&P 12 months' earnings. Combined with wide month-to-month swings, P/E projections based on analysts' earnings predictions proved unreliable.

Sobering Picture

Gauged by several ratios, the stock market looks pricey, compared with its valuation during other periods.

Figure 1 shows the comparison of S&P projected earnings to actual earnings over a 298-month period from 1979 through 2003.

Figure 2 shows the price-to-earnings ratio of U.S. nonfinancial stocks, based on actual company profits, as reported by the Federal Reserve.

Today's S&P price-to-earnings ratio of 14, based on Federal Reserve data, is well below its peak of 51, hit in 2001, but is still historically high.

Comparing today's price-to-earnings ratio to those from 1945 through the early 1980s, it's easy to understand why some analysts were concerned about overvaluation soon after the current bull market started in 1982. The question that investors must answer is: What happens if the price-earnings ratio reverts to its pre-1982 range?

Figure 3 shows a semi-log chart of the price-to-book ratio of U.S. nonfinancial stocks since 1945, as reported by the Federal Reserve.

This price-to-book ratio compares the end-of-quarter total dollar value of all U.S. nonfinancial stocks to the Fed's tabulation of the dollar value of company assets for the same periods. It is a simplified version of James Tobin's "q ratio," comparing the stock-market value to the asset value of U.S. companies. While below the 2000 peak levels, the ratio is currently above its level from 1945 through 1994.

Of the three ratios, the price-to-book ratio has historically been the most reliable in valuing the stock market. But at the 2009 stock market lows, the ratio dropped only to its long-term average level. The danger is that this ratio could fall past its long-term average level toward its past historic low, and that would bring a further loss of as much as two-thirds of its current value.

Figure 4 shows the price-to-dividends ratio of the S&P 500 since 1945, computed from Shiller's data.

Before the unprecedented tech boom of the 1990s, investors received a dollar in dividends, on average, at bull-market peaks for each $33.80 worth of the S&P index they owned. At the 2000 peak, an investor needed to own $90.20 of the index to get $1.00 in dividends. Currently, with the S&P around 1200 and a dividend of $23.19, an investor needs $51.75 of the benchmark index to get $1.00 in dividends. While far below 2000 peak levels, the price-to-dividends ratio now is higher than it was in every month from 1872 through 1996.

Taken together, these ratios present a time-tested picture of stock-market valuation levels that an investor should consider when deciding what portion of his or her investments should be allocated to stocks. It's impossible to know if the elevated levels of these three ratios over the past 20 years represent a new normal or an extended bubble that will eventually pop.

Still, investors should ask: "What if this time it's not different, and valuation levels revert to the ranges that prevailed before 1982?"