Regardless of your opinion of President Bush’s proposal to privatize part of the Social Security system, I think we can all agree that the debate over that proposal should be conducted with realistic assumptions about what returns the stock market will be able to produce over the next several decades.

This does not appear to be the case currently. Many participants in the debate are blithely assuming that stocks in the future will be able to match the 7% real, or inflation-adjusted, return that they produced over the last 200 years.

That assumption is way too high, according to Robert Arnott, chairman of Research Affiliates, a research and asset management firm based in Pasadena, California and Peter L. Bernstein, president of Peter L. Bernstein, Inc., a New York-based institutional consulting firm. In an article they wrote for the March / April 2002 issue of the Financial Analysts Journal, they pointed out that much of equities’ historical return has come from what they call revaluation -- investor optimism that leads the market to place a higher price on a dollar’s worth of earnings.

Because investors can just as easily decide to place a lower price on a dollar’s worth of earnings as a higher price, it would certainly seem to be dangerous to base retirement expectations on the assumption that investors will continue to place higher and higher valuations on equities in the future.

John Maynard Keynes said as much nearly a century ago when he wrote that “trees don’t grow to the sky.” Yet, in effect, what investors are assuming will happen, when they extrapolate past returns into the future, is the economic equivalent of trees growing to the sky.

Without revaluation, what can we expect for stocks in the future? According to Arnott and Bernstein, stocks’ returns have just two components other than revaluation: Dividend yield and real growth in earnings. And, these two other factors today do not justify a return expectation of 7% above inflation.

Arnott and Bernstein found from their studies that, since 1871, earnings have grown at an inflation-adjusted rate of approximtely 1.5% annualized. Couple that with the market’s current dividend yield, also approximately 1.5%, and we arrive at an expected real return of just 3 percent annualized.

To justify a return expectation of 7%, therefore, one needs to look to revaluation to add four percent. Yet, that is exceedingly unlikely, according to Arnott and Bernstein. Stocks’ dividend yield is already much lower than its historical average, and the market’s current P/E ratio is higher. It, therefore, is more likely than not that the market’s future dividend yield will be higher than today’s, and its P/E ratio will be lower.

That prospect is very sobering. To the extent that the market’s future yield is higher than today’s and its P/E ratio lower, the market’s expected return is even less than 3% per year. In other words, even assuming just a 3% real return may be too optimistic.

Isn’t all this pessimism simply a function of the market’s current high valuation? Wouldn’t a real return assumption of 7% be rational in more normal circumstances?

No, according to Arnott.

Consider a severe bear market in which increased investor pessimism forces the market’s P/E ratio back down to its historical average and doubles the market’s dividend yield to 3 percent. Note that, even so, the market’s yield would still be below its historical average. Even with such lower valuations, according to Arnott, it would still be irrational to expect revaluation to provide a boost to equity returns.

That’s because, at such a point, it would be just as likely for future valuations to be lower as it would for them to be higher -- and it would be dangerous to bet one's retirement on the latter.

What would a realistic expectation be for stocks’ long-term real return following such a bear market? According to Arnott, it would be 4.5 percent per year -- 3 percent from dividend yield and 1.5 percent from earnings growth. That’s still 2.5 percentage points per year less than the 7% rate that many are assuming in the current social security debate.

Note carefully that adopting a realistic return assumption has nothing to do with how you come down with respect to whether privatizing Social Security is a good idea. In an email message earlier this week, for example, Arnott wrote that, despite his argument in favor of lower return assumptions, he still “strongly favor[s] privatization, but not because it’ll ‘save Social Security.’ Nothing short of a substantial increase in the normal retirement age will do that. I favor privatization because it moves us one small step away from the pervasive ‘nanny state’ mentality, in which we look to big brother to take care of us. Big brother won’t take care of us; it’s time for us, as citizens, to start to take more responsibility for ourselves.”

Finally, I should note that participants in the Social Security debate are not the only ones with unrealistic return expectations. Corporations with defined-benefit pension plans also fit that category. In an early-January column for the New York Times, I discussed the implications of Arnott’s and Bernstein’s argument for those corporations’ earnings. Those implications are no prettier for them than for privatized Social Security accounts.

Mark Hulbert is editor of the Hulbert Financial Digest, a service of Marketwatch that, for nearly 24 years, has tracked the performance of investment advisory newsletters. A section of the Marketwatch website called "Hulbert Interactive" (marketwatch.com/hulbertinteractive) allows users to conduct extensive research on the HFD database.

Mark can be contacted via email at mhulbert@marketwatch.com