There are two approaches to questions of earning power, the first, annual, which is the one adopted in this paper, the second, a whole-life study. Of the two the latter is incomparably superior. A study of annual rates of earnings is inevitably biased by the fact that it can include the surviving corporations only. Those which fail make no reports, nor are the capital losses in failure offset against the profits of the survivors to show the net return to capitalists as a class upon their ventures. In addition, the rates of earnings of any year are affected by the business conditions of that year. It is only when one has a series extending over a full business cycle that one has material from which to compute a “normal” rate of return, subject to the inclusion therein of a bias which is not open to any reasonably dependable estimate.

In the whole-life study the rate of return found is that rate which will equate the in-payments of capital by investors to the corporation with the dividends paid to investors including therein the liquidating dividend(s) if the corporation is wound up (or the current market value as a reasonable substitute therefor if it still continues). The beauty of this method is that it avoids the whole problem of allocating earnings to particular time-periods and escapes the illogicality of showing as earnings what never reaches the owner. Its disadvantage is that it is applicable to long periods only. A decade is probably the minimum and two are preferable. It is the only method of determining earning power which is logically defensible; but for many purposes the annual data, with all their imperfections, are still useful and it is for such uses that this study is presented.