Various studies have investigated variation in reporting conservatism with pre-specified contractual incentives. Lawrence et al. (forthcoming), hereafter LSS, propose a model to control for “normal” or “non-discretionary” conservatism while testing for variation in conservatism with contractual forces, which they characterize as “discretionary” variation. Our objective is to further the discussion on discretion in conservatism by relying on LSS and incorporating insights from related studies to shed light on the rationales and relative importance of various controls for normal conservatism suggested by LSS. Furthermore, our perspective on discretion in conservatism is distinct from the one developed by LSS in some crucial respects, ultimately leading us to question the nature of the boundary between discretionary and non-discretionary conservatism.

We investigate whether insider selling affects the likelihood of firms receiving auditor going-concern opinions. Prior studies document significant negative market reactions to the issuance of going-concern opinions, indicating that such opinions convey bad news to investors. Insider sales followed by negative news are likely to attract regulators' scrutiny and investor class-action lawsuits. Therefore, we predict that, to reduce the risk of litigation, managers have incentives to avoid receiving going-concern opinions after their insider sales by pressuring auditors for clean audit opinions. We evaluate this prediction empirically and find that the probability of receiving a going-concern opinion is negatively associated with the level of insider selling. Further analysis indicates that this negative relation is more pronounced for firms that are economically significant to their auditors but less pronounced when (1) auditors have concerns about litigation exposure and reputation loss and (2) audit committees are more independent. Finally, the negative relation between going-concern opinions and insider sales is significantly weakened after SOX.