What can we conclude from common tests of accrual mispricing?

Mispricing and risk have both been suggested as explanations for the cross-sectional relation between stock returns and firm characteristics such as accruals. As emphasized by Ferson and Harvey (1998) and Berk, Green and Naik (1999), it is difficult to evaluate these competing explanations without explicitly modelling the relation between risk and firm characteristics, if risk is not independent of firm characteristics. Drawing on theory and empirical evidence, this paper models systematic risk as a function of accruals, and accruals as mean-reverting. When the true abnormal returns are zero, but the true betas are unobserved by the researcher, the model predicts the anomalous pattern of empirical results previously reported in the accrual anomaly literature and attributed to mispricing: (i) CAPM abnormal returns to an accrual hedge portfolio are positive on average; (ii) the abnormal returns are positive in almost all years; (iii) abnormal returns decay as the holding period is extended beyond one year; (iv) the Mishkin (1983) test of market efficiency is rejected. Using simulations, the paper shows that small and plausible degrees of risk mismeasurement also reproduce the magnitudes of the anomalous results previously reported in the literature