In this study, we present a nonstrategic, dynamic Bayesian model in which auditors' learning on the job and their choice of professional services jointly affect audit quality. While performing audits over time, auditors accumulate client‐specific knowledge so that their posterior beliefs about clients are updated and become more precise (that is, precision is our surrogate for audit quality) — what we call the learning effect. In addition, auditors can enrich their knowledge accumulation by performing nonaudit services (NAS) that, in fact, may influence clients' managerial decisions — what we call the business advisory effect. This advisory effect permits auditors to anticipate and to learn about changes in clients' business models, which in turn improves their advisory capacity. These dual “learning” and “advisory” effects are interdependent and mutually reinforcing. The advisory effect of NAS may increase or reduce auditors' engagement risk. We show that large professional fees can induce auditors to provide NAS that increase engagement risk and diminish audit quality. However, when NAS reduce engagement risk and increase audit quality, auditors may provide NAS without charging clients. The feature that distinguishes our study — the interdependence between the learning and advisory effects — provides new insight into the trade‐off between audit fees and audit quality. Consequently, our analysis helps explain why the scope of the audit has evolved over time and why the boundaries between audit and NAS are constantly shifting. A recent example of such a shift is that the Sarbanes‐Oxley Act adds control attestation to audits for public companies traded in U.S. markets.

We examine whether the provision of nonaudit services (NAS) by incumbent auditors is associated with a reduction in the extent to which earnings reflect bad news on a timely basis (that is, news‐based conservatism). Reduced conservatism is expected to occur if relatively high levels of NAS result in reduced auditor independence and, ultimately, lower‐quality auditing. Because client‐specific demand for NAS is expected to vary, our proxy for the auditor‐client economic bond is the extent to which NAS purchases (relative to audit fees) are greater or less than expected. Using several different methods for identifying news‐based conservatism, we consistently find that higher than expected levels of NAS are not associated with reduced conservatism. This result is robust to allowing for endogenous NAS demand, as well as several explicit factors that may be associated with differences in conservatism. Similar conclusions arise from tests that use alternative measures of the economic bond between auditors and their clients, as well as in tests confined to either the Big 6 or non‐Big 6 audit firms. Our results are consistent with factors such as market‐based incentives, the threat of litigation, and alternative governance mechanisms offsetting any expected benefits to the audit firm from reducing its independence. We therefore conclude that recent legislative intervention aimed at restricting the supply of NAS is unlikely to result in increased independence in fact, although independence in appearance may be improved.

We examine the determinants of managers' use of discretion over employee stock option (ESO) valuation‐model inputs that determine ESO fair values. We also explore the consequences of such discretion. Firms exercise considerable discretion over all model inputs, and this discretion results in material differences in ESO fair‐value estimates. Contrary to conventional wisdom, we find that a large proportion of firms exercise value‐increasing discretion. Importantly, we find that using discretion improves predictive accuracy for about half of our sample firms. Moreover, we find that both opportunistic and informational managerial incentives together explain the accuracy of firms' ESO fair‐value estimates. Partitioning on the direction of discretion improves our understanding of managerial incentives. Our analysis confirms that financial statement readers can use mandated contextual disclosures to construct powerful ex ante predictions of ex post accuracy.