The purpose of this research is to test the effects of agency incentives and knowledge spillovers in joint engagements for audit and nonaudit services. External parties and regulators have associated joint nonaudit purchases with beliefs about impaired auditor independence. Since reductions in perceived independence correspondingly reduce audit credibility, agency theory suggests that auditees have an incentive to limit nonaudit purchases because of the agency costs imposed when an audit's monitoring value is diminished. The frequency of nonaudit services is expected to reflect knowledge spillovers from joint engagements and the level of economic bonding between the auditor and the auditee. Hence, two categories of services are examined, recurring and nonrecurring. Agency incentives are expected to significantly influence the auditee's decision to purchase recurring nonaudit services since the auditor's revenue from a recurring service can be perceived as an annuity. However, for a nonrecurring service, future knowledge spillovers and incremental economic bonding are expected to be minimal. Because the auditor's revenue from the nonrecurring service is not an additional annuity, any perceived impairment to independence and credibility caused by incremental economic bonding should be limited. Data from Fortune 500 companies, disclosed under Accounting Series Release (ASR) No. 250, are used in multiple regressions to test whether auditees vary nonaudit purchases in accordance with the potential agency costs that could arise. The findings indicate that variables related to expected agency costs (management ownership, outside investment concentration, and leverage) significantly explain cross-sectional differences in the demand for recurring nonaudit services. The data also indicate that auditees purchase higher levels of recurring services when they engage particular auditors (i.e., industry specialists or the audit firm that is the largest overall provider of nonaudit services). Consistent with expectations, overall agency costs do not explain the level of nonrecurring services purchased from the auditor. This latter finding suggests that selection of an industry specialist may be a quality signal that allows an auditee to purchase higher levels of recurring nonaudit services than would normally be expected. Future research may explore how nonaudit services influence auditor switching. For example, environmental factors could increase the need for nonaudit services, thus making it more efficient to change to an industry specialist. Alternatively, a reorganization of ownership or capital structure could reduce the need for services, which would lead to a renegotiation of an existing joint services contract. At a minimum, the results reported here suggest that auditees evaluate a potential auditor's ability to provide a particular type or level of nonaudit services.
