With the strong possibility of a recession looming and stock prices sliding like cars on an icy road, investors have to look to the downside. How bad can it get for corporate earnings?

As earnings power is likely to drop after years of remarkable growth, it is time to check out potential bad scenarios.

One growth statistic stands out: U.S. corporate profits, as a proportion of gross domestic product, doubled to a record 14% in 2006 from 7% in 1982. As we know, Canadian economic trends follow the U.S. pattern. Corporate return on equity in Canada reached 14% in 2006, up from less than 2% in 1992, according to Statistics Canada.

After gains like these, a reaction is more than likely.

Let’s avoid the thinking described by Benjamin Graham and David Dodd in Security Analysis: “The investment community nearly always expects that the general experience of the past decade will be continued into the next, and in this it has virtually always been wrong.”

When economic prospects appear favourable, latest three-year average earnings and profitability make a solid foundation on which to base estimates for the future.

But when a contraction looms, those averages are not the best tools to use. Current circumstances — namely, the long business expansion that’s now ending — suggest a long look back when you study investment prospects.

If a recession — or a severe slowdown — occurs, revenue will drop; and falling with it will be profitability. But how far is down?

The first possibility in a bad year is that a company’s profitability will approximate the average of the past 10 years. This time span encompasses all changes likely to influence current operations. As an alternative, a seven-year look back would also be useful.

The worst-case scenario is taking the average of a company’s three worst years in the past 10, excluding years in which a loss was recorded. If a company manages to be profitable, we want an idea of how profitable it might be under stress. After all, a loss is a loss.

The accompanying table provides a sample of what you would find using this strategy. The 11 companies, taken at random from the Canadian market, represent a variety of industries and corporate sizes. Companies in the sample with losses in the 10 years covered include TD Bank Financial Group, Agrium Inc., Le Château Inc. and Héroux-Devtek Inc.

The profitability measure used here is average return on equity. Return on invested capital is an alternative measurement.

Similarly, compare the 10-year average earnings per share to recent results for another slant on possible downside earnings. For the worst-case scenario, compare average earnings for the three lowest years — again, excluding loss years.

The unknown in this exercise is what multiple the stock market will be willing to pay for reduced earnings for any company. It will pay to be conservative in your estimates, and expect multiples to drop as the market continues to decline.

The next step is to review the pattern of the past 10 years. An average can provide a trend, but it can also hide information. Teck Cominco Ltd. is a resources sector example. Its 10-year average ROE is 10.5%. This obscures huge changes in that period. The company’s return jumped from single digits before 2004, when the industrial metals boom resulted in a jump to more than 20% ROE in the next three years. This is no surprise because we know resources industries are a feast or famine affair. The numbers indicate how severe and fast the change can be.

In situations for which there has been a major shift in operations, as with Jean Coutu Group, another detailed review is needed. With the sale of its U.S. retail drugstore operations, Coutu’s ROE dropped abruptly — even though earnings have been relatively stable.

To gauge the direction in which a company’s earnings might go in a business slowdown, you would apply the average ROE to current shareholders’ equity.

For example, the 10-year average ROE for Fortis Inc. applied to the shareholders’ equity reported at its latest fiscal yearend produces theoretical earnings of $151 million. This compares with actual fiscal 2006 earnings of $147 million. Should Fortis’s ROE drop to the average of the three worst years in the past 10, it would earn $122 million.

@page_break@A similar projection for Power Corp. would produce theoretical earnings of $1.4 billion on its 10-year average ROE and $1.1 billion for its worst three years, compared with fiscal 2006 earnings of $1.4 billion.

A note on the data used: earnings exclude extraordinary items such as goodwill write-offs, as we are interested in operating results. Because corporate reports for 2007 are just coming out, the data series end with 2006 or with “fiscal 2007” figures for companies whose years ended early in 2007. IE