Stocks have swung so wildly over the past two years that their prices must seem to non-investors to be the product of whimsy rather than fundamentals like company earnings. The S&P 500 index, a broad basket of American shares, topped 1576 in October 2007 and plunged below 667 in March. Recently, it has loitered nervously near 1000.

Assuming the cold calculus of price/earnings ratios will soon win out over nerves in deciding the direction of shares, investors should consider what today s P/E for the S&P 500 says about the future. Unfortunately, the numbers aren t comforting.

Over the past 136 years, stocks have traded at an average price of just under 15 times trailing earnings. With the S&P 500 index at 1000, stocks are nearly at 25 times trailing earnings. (I m including earnings for the second quarter of 2009 in the trailing number because most companies have already reported results for it.) So stocks look 67% pricier than usual.

Factor in Wall Street analysts' growth estimates for future earnings and there s good news. Earnings over the next four quarters are forecast to jump 62%. By that view, stocks aren t terribly overpriced. They re just waiting for profits to catch up.

There s just one problem. Analysts have historically demonstrated little ability to accurately predict earnings for the current quarter, to say nothing of the three quarters after that.

There is a more reliable way to adjust trailing earnings -- one that doesn't rely on any analyst guesswork. In their book Security Analysis (1934), Benjamin Graham and David Dodd recommend adjusting for swings in the business cycle by using a trailing 10-year average of earnings when calculating the market s P/E. By my math, that puts the S&P 500 at 16.4 times earnings. Yale economist Robert Schiller, a proponent of the 10-year P/E, calculates his own version and on Aug. 3 got 17.6. Based on those measures, the market is only moderately expensive -- somewhere between 9% and 17% pricier than average.

However, the 10-year P/E could mislead if the prior 10 years produced skewed numbers. That is, if we re coming out of 10 bubbly years, an average based on those years probably won t reflect normality.

Have a look at the two tables below. Both show U.S. after-tax corporate profits as a percentage of national income -- the country s profit margin, if you like -- as reported by the Bureau of Economic Analysis. The first chart includes numbers starting in 1999 and the second starts in 1929. Both end after the first quarter of 2009, when the nation s profit margin was 7.5%. But the 10-year chart makes it look as though that level of profitability is below average, while the longer-term chart shows clearly that profits are, if anything, above average.

Even if we erase the effects of the Great Depression by starting the numbers at 1940, we get an average profit margin of 6.7%. Profits, in other words, might be 12% too high. That might seem counterintuitive given the dramatic slowing of consumer spending over the past year. In June, Americans saved 4.6% of their discretionary income up from less than 1% over four years ended 2008. But again, recent history misleads. The average savings rate going back to 1929 is 7.4%. American s are spending sharply less than in recent years but well more than the historical average.

If we adjust the S&P 500 s trailing P/E ratio to account for the nation s above-average profit margin, stocks are perhaps 28 times earnings or 87% overpriced.

There s one more thing. Until now, I ve been using a measure called operating earnings. It ignores certain non-recurring transactions like write-offs of bad debt and gains from asset sales. It s common practice today to ignore such things based on the view that they don t relate to ongoing business. But write-offs relate to a company s prudence in issuing credit and the outcome of asset sales speak of the wisdom of past investments. Until fairly recently, investors counted everything when examining earnings. The 136-year P/E ratio I gave earlier (about 15) is based on as-reported earnings, warts and all, not prettied-up numbers that show what companies would have earned if not for this or that special event.

Besides, increasingly, the events aren t so special. Standard & Poor s publishes operating earnings and old-school as-reported earnings for its 500 index going back to 1988. In the late 1980s and early 1990s, operating earnings were sometimes less than as-reported numbers. Then companies got wise to the practice of trumpeting operating earnings to investors while quietly tucking bad news into as-reported earnings. For the past 14 years, operating earnings have been higher than as-reported earnings every quarter.

Much as I d like to, I can t use as-reported earnings for the market s P/E. So devastating were bank losses over the past year that the sliver of profit left for the market would put shares at more than 130 times earnings. But if we want to make a fair comparison with the historic P/E for stocks, we should discount those operating earnings to bring them in line with what earnings used to be. Cumulatively since 1988, operating earnings have been 19% higher than as-reported earnings. If we discount trailing earnings by that much, the market s P/E rises from 28 to about 35 -- yikes.

Keep in mind that in fiddling with trailing earnings, I ve done only two things: adjusted for where the nation s profit margin stands relative to its long-term average and adjusted them again to bridge the gap between dolled-up earnings and real ones. If I ve been overly cautious -- if we assume, for example, that the nation s higher profit margin represents a new normal and that operating earnings are, despite appearances, on the level -- we re still back to 25 times trailing earnings. That's still two-thirds pricier than average. Meanwhile, high prices have shrunk the S&P 500 s dividend yield to 2.2%, less than half the average for stocks over the past two centuries.

Stock prices can easily rise from here, of course. The economy might expand faster than expected, inflation might devalue the money relative to all assets including stocks, or investors might simply follow one another into the decade s third stock bubble. But prudent investors should raise cash and be highly selective in their purchases.