This study makes two main contributions to the literature. Firstly, it tests empirically the relative timeliness of accrual measures and earnings components used as explanatory variables in accrual models (“accrual drivers”) regarding the impact of conservatism. Secondly, taking into account the empirical evidence on such a timeliness, it discusses intuitively potential implications for the specification of (traditional) accrual models and the quality of discretionary accrual estimates. It concludes that common accrual models, as Jones (1991), are misspecified. They have a dependent variable (accruals) asymmetrically affected by conservatism, and one or more explanatory variables that are not affected, inducing a non-systematic measurement error in estimating discretionary accruals.