Share Price Changes and Price/Earnings Ratios as Predictors of Fraud Prior to a Fraud Announcement

Rapidly changing economic environments led to new methods of committing financial fraud. While fraud indicators that require non-public corporate information were previously identified, research on fraud indicators using publicly available information (such as share price) was absent. According to the efficient market hypothesis, all information, both public and private, is included in share price. Thus, corporate share price should reflect fraud prior to a public fraud announcement. The purpose of this quantitative study was to test the efficient market hypothesis by determining the extent to which changes in share price and price/earnings (P/E) ratios prior to a public announcement of fraud predicted whether a company was prosecuted for fraud. An ex post facto, secondary data analysis was conducted using the coefficient of variation and the price/earnings ratio as the predictor variables and fraud status as the criterion variable. Data was collected from 139 companies listed with the SEC and traded on an American stock exchange between 2000 and 2004. Companies convicted of fraud were matched with companies of a similar size within the same industry that were not convicted of fraud over the same time period. Data was entered into a logistic regression to determine whether changes in share price predicted whether or not the company was prosecuted for fraud. Results revealed a significant relationship between companies prosecuted for fraud and the coefficient of variation (Wald[1] = 4.6,p = .03). However, the relationship between the price/earnings ratio and companies prosecuted for fraud was insignificant (Wald[1] = 0.99,p = .32). Thus, the strong-form version of the efficient market hypothesis was supported only through the coefficient of variation. Results from this study support the use of quantitative measures that can help stakeholders detect fraud early to minimize costs. Further research can build on these fmdings to create a more robust model with greater accuracy for detecting fraud.