Commentary

Don't Sell Out

By Burton G. Malkiel. Mr. Malkiel, a professor of economics at Princeton, is author of "A Random Walk Down Wall Street" (W.W. Norton, 7th edition, 2000).

The horrific events of late have focused the attention of the world on the U.S. stock market and the potential influence it will have on the likelihood of a global recession. As of yesterday's close, the Dow is more than 10% below its level prior to the terrorist attacks and consumer confidence has collapsed.

While more than a $1 trillion loss in market value was punishing enough, many commentators have expressed concern that the market is still overpriced and that the potential for further declines is large. Simply put, the worry-warts argue that with today's price/earnings ratio (P/E) still in the low 20s (well above its long-run average of about 15), stocks are overpriced and that corporate earnings are greatly overstated, making the actual P/E dangerously high.

While stocks may continue to fall in the short-run, investors should reject such a negative view.

Correct Earnings Multiple

There is no reason to believe that 15 or any other number is the correct earnings multiple at which the market should sell. Two important factors influence the ratio. The first is the level of long-term interest rates. In the early 1980s, when long-term government bonds had generous double-digit yields, stocks had to fall relative to earnings to provide competitive expected returns. Thus, the price/earnings multiple for the market at that time had to be below average. And when long-term interest rates are very low, as they are today, higher P/E ratios for the market are warranted.

The appropriate P/E for the market also depends on the risk premium (the extra return over safe bonds) demanded by market participants. One very good proxy for the risk perceived by equity investors is the recent inflation rate. High rates of inflation augment the perception of risks for two reasons. First, inflation tends to make investment planning more difficult and considerably riskier. And second, inflation usually increases risks throughout the economy. High rates of inflation tend to induce the Federal Reserve to "take the punch bowl away" to keep the party from getting completely out of hand. Inflation then increases the risk that the economy will slow in the future and weaken corporate profits. When inflation rates are high, P/E multiples for the market should be low.

I have analyzed monthly data spanning the past 35 years. It shows the actual P/E for the Standard & Poor's 500 index and the P/E that would be predicted based on the past 12 months' change in the Consumer Price Index. The P/Es bear a close relationship to bond yields and the rate of inflation. The normal relationship today, based on bond yields and inflation, is for the S&P to sell at about 22, not 15. Bond yields are low and inflation is well contained, so it is perfectly appropriate for the S&P to sell at a level above its long-run average. The market today appears to be fairly valued.

But do the tragic events of Sept. 11 throw all the historical analogies out the window? I think not. I have looked at previous shocks to the economy: the Gulf War, the crash of 1987, the resignation of President Nixon, and the U.S. bombing of Cambodia in 1970, among others. While one can discern temporary market perturbations, there do not seem to be any long-run deviations from the prediction line. Even if I extend the analysis back in time to the Korean War and to Pearl Harbor, I find only temporary effects. Of course, it is always possible that this time is different, but history suggests that investors who make an emotional decision to sell during times of crisis are unlikely to derive any benefit.

My analysis assumes that the earnings numbers are reliable, and that the earning power of U.S. corporations will not be impaired seriously by the economic fallout from the terrorist attacks and the retaliation to come. Yes, earnings are likely to decline in the coming quarters, but there has always been a tendency for P/E multiples to rise during recessionary periods. And even if we do have a recession, the market is likely to look beyond depressed current earnings and to capitalize earning power during the following recovery.

Critics argue that the earnings we use today to measure the earnings per share of the Standard & Poor's 500 companies have, increasingly, become less meaningful. Many companies today focus investors' attention on "pro forma" earnings, where no rules or guidelines exist, rather than on earnings calculated under generally accepted accounting principles. Pro forma earnings are often called "earnings before all the bad stuff," and give firms license to exclude expenses they deem to be "special," "extraordinary" and "nonrecurring."

Depending on what expenses are considered to be improperly ignored, different analysts have documented a substantial overstatement to reported earnings. The Wall Street Journal has suggested that the "true" P/E multiple for the S&P 500 is in the mid-30s rather than the low 20s. Goldman Sachs has estimated that "pro forma" S&P earnings per share may be $10 too high. Bernstein Research suggests that the net profit margin for industrial firms was nearer 5% than the 7% reported in 2000.

Cyclical Changes

There is no doubt that legitimate issues can be raised about the quality of reported corporate earnings, and that quality has most likely deteriorated in recent years. But the real issue is whether there has been deterioration over the long term. Here, I am convinced that the changes in earnings quality are cyclical, and that far greater concerns could be raised about earlier periods than those being raised now.

In the inflationary early 1970s, for example, reported earnings were seriously overstated by the inclusion of nonrecurring inventory profits and from depreciation charges that were too low. And "nonrecurring" write-offs were several times larger in the early 1990s than they were in 2000 and 2001. Thus, by some measures, earnings quality may well be better today than in the 1970s and early 1990s.

While the market today seems reasonably priced, no one can make a short-term prediction of where prices will go in the future. Certainly the world is a much less stable place than it was before Sept. 11, and risk premiums should be higher. Moreover, if irrational exuberance characterized 1999 and early 2000, unreasonable anxiety could influence prices over the next several months. Indeed, figures released yesterday showed the biggest monthly drop in the consumer confidence index since 1990. But history tells us that anyone who sells stocks today in the hope of getting back in at just the right time is likely to be making a large and costly mistake.