Research Design Issues and Related Inference Problems Underlying Tests of the Market Pricing of Accounting Information

Abstract

Numerous accounting studies conduct tests of the market pricing of accounting information. The purpose of this study is to both highlight and quantify the consequences of using of ex-post information to form trading strategies based on accounting numbers and to document the importance of performing robustness tests (such as a careful analysis of outliers), when testing economic or behavioral explanations for apparent accounting-related (mis)pricing. To illustrate the importance of these issues we focus on the accrual anomaly (see Sloan 1996). We show that correcting for common sample selection biases not only reduces the magnitude of the abnormal returns to the hedge portfolio, but also dramatically alters the abnormal returns attributable to the high and low accrual portfolios. After eliminating outliers we document an inverted U-shaped relation between buy-and-hold abnormal returns and accruals. This result, combined with a detailed examination of both the truncated sample and the outliers, leads us to conclude that it is neither investors' inability to process accounting information (Sloan 1996) nor their excessive optimism with respect to glamour stocks (Desai, et al 2004) that explains the accrual and operating cash flow anomalies.