I recently read this in the Financial Times, and I loved it:

The rating of the [Nifty Fifty] reached its peak in 1972, when shares in Coca-Cola and Xerox each cost 46 years' earnings. Over the following 25 years, the ratings fell to Earth, while the actual performance of the companies allowed Wharton School professor Jeremy Siegel to work out whether those eye-watering ratings had been justified. Some had turned out to be absolute dogs: Xerox's performance warranted a price/earnings ratio of 18 rather than 46. But Coca-Cola was so successful it would have been worth paying 92 times 1972 earnings.

That's a neat way to think about it. How much should you have been willing to pay for a company in hindsight to achieve an acceptable rate of return? Put differently, what P/E ratio could I have purchased a company for in the past and still earned an average rate of return through today?

I wondered how this logic would look if we applied it to today's Dow stocks. So here's that, using 1995 as a starting year and requiring an 8% annual return (including dividends): 

Sources: S&P Capital IQ, author's calculations, and an afternoon wasted in an Excel spreadsheet.

It's a huge range. UnitedHealth Group may have looked pricey at 28 times earnings in 1995, but you could have paid three times that amount and still earned an 8% annual return over the following 17 years. Bank of America (NYSE:BAC) may have looked cheap at seven times earnings, but it was worth only a fraction of that if you held through today.

This is merely a hindsight exercise, and doesn't really say what you should have done in 1995. No one could have known in 1995 that Bank of America would go hog wild during a mid-2000s housing boom and dilute its shareholders down to the ground in the aftermath. And no one could have known that McDonald's would successfully navigate a huge international expansion. Investing is about risks and unknowns, which have to be accounted for in current prices. The fact that we don't know what the future outcome will be is what allows stocks to earn higher long-term returns than bonds or cash.

But what strikes me about this table is how wide the range is. The average absolute value of the difference between the actual 1995 P/E ratio and the P/E ratio you could have rationally paid is 18.1 -- a huge number if you think about it.

To me, the takeaway is that over long periods, most companies will either be very successful or very disappointing, with less in between than most of us probably imagine.

What do you think?