A Comparison of Measures of Earnings Per Share

This paper explores alternative methods for computing earnings per share (EPS) for a company whose capital structure consists of ordinary shares and warrants. The methods for computing EPS identified by the FASB (1996) are critically evaluated and an alternative measure, the holding period approach,is developed within the framework of contingent claims analysis. Two types of errors are shown to characterize the accounting measures of EPS. One arises from failure of accounting measures to fully recognize the contingent nature of the warrant. The other arises from the practice of not recognizing instances of anti-dilution. A further factor is the treatment of any difference between the proceeds from the issue of the warrants and their fair value at that time. This is ignored in existing measures and yet may have a significant effect on the value of the claims of ordinary shareholders on the company's earnings. Using a simulation method it is shown that the imputed earnings method of computing EPS is a very close approximation to the holding period method and is considerably more accurate than treasury stock measures favoured by accounting standards bodies.