Projecting earnings growth is a risky business at the best of times. Accurate estimates even a year or two ahead are uncommon — there are just too many unknowns in the equation. It is especially tricky now, when there is such uncertainty about business conditions.

The best thing to do, under the circumstances, is estimate what a company can earn in the future based on its past average earnings. Three years is a preferred time frame, one used by the father of investment analysis, the late Benjamin Graham.

Average per-share earnings of past periods are indicators of future earnings, Graham and his associates argued in their classic textbook on security analysis. It is a safer starting point for valuing a stock than just one year’s results. The three-year period picks up recent changes in a company’s operations, eliminating influences now out of the picture. Some analysts, however, prefer a five-year average for judging what a business can do in an average year.

Using the latest five-year period for Canadian companies and stock indices sharply cuts average earnings figures from those shown in the table at right. Reason: there has been rapid growth in earnings since 2002. In that year, the S&P/TSX composite index recorded a loss instead of a profit.

Whether the latest three-year average earnings are “normal” is another question, the answer to which depends on individual corporate and industry factors. But such an average does show what a company or industry has done, and could do again.

With average earnings as a firm base for estimating the value of a stock, you can decide whether earnings are trending upward and, if so, at what rate, and then put the two together.

You can also assume the expected price/earnings multiple. With P/Es added into the equation, you can decide whether a stock at its current price is worth buying — or selling.

In the accompanying table, broad industry sectors of the stock markets in the U.S. and Canada are the subject of the analysis. The Canadian indices are capped. We have calculated the latest three-year average earnings to March for the Standard & Poor’s 500 composite index and to June for the S&P/TSX composite.

Another column in the table projects earnings of each sector using a 20% increase.

P/E multiples are all over the map for both three-year average earnings and for the projected 20% higher earnings. The current P/E on average earnings for U.S. sectors ranges from around 12 to as high as 50. In Canada, the range is even wider.

What P/E should we assume for the future? There is always controversy about this, as was explained recently by U.S. investment consultant Richard Daughty, who writes the Mogambo Guru newsletter.

Writes Daughty: “A P/E of 20 means that you are spending 20 bucks to buy a share of stock in a company that makes one lousy dollar per year per share! In other words, the company itself will not earn enough to equal what you paid for it for [after] 20 long, long years of waiting and hoping.

“This is why this ‘P/E of 20’ thing is around the point at which all previous stock markets, both on this planet you call Earth and on all the other planets in the cosmos, both now and in all of history, eventually fell, corrected or collapsed and ruined the hell out of everything,” he writes. “In case you were wondering, the historical average P/E for a stock or stock market is around 12 to 14 or so, although it has gotten down to around four to seven or so at big market bottoms and around 20 at the tops of bull markets.”

Canadian experience with P/Es is, of course, much more volatile. As mentioned, the S&P/TSX composite had a P/E of zero a few years ago. It has also sported a P/E of 100 at times. Currently, its P/E is 17.

Leaning on the optimistic side, the table assumes average earnings now and that the projected 20% higher earnings should be valued at 15 times earnings. The table then shows how much the current stock price overvalues or undervalues these notional valuations.

For instance, the S&P 500 is trading 14% higher than three-year average earnings valued at 15 times P/E. If earnings grow by 20% and the market in the future values them at 15 times, there is a 3% upside for the S&P index. On the same basis, the TSX composite is currently 17% higher than the projected valuation.

@page_break@TSX composite earnings are up 50% in three years, and up from less than zero in five years. S&P 500 earnings are up 28% in three years and up 87% in five years. With this in mind, and if you accept the assumption that 15 will be a reasonable future P/E for any stock sector, you get the impression most market sectors are more than fully valued.

To assume otherwise, you must expect continued fast earnings growth and higher P/Es. IE