In this article, we use audit-hour and billing-rate data supplied by a large public accounting firm to address the question, ''Does providing audit clients with nonaudit services result in knowledge spillovers and audit production efficiencies that could produce economic rents for the auditor?'' In prior analytical work, both Simunic (1984) and Beck et al. (1988) have argued that knowledge acquired while providing nonaudit services may ''spill over'' to the production of the audit, and thus generate production efficiencies. If audit production efficiencies lead to cost savings that are retained in whole or in part by the auditor (rather than passed on to the client), then economic rents accrue to the auditor, creating incentives for the auditor to resolve disputes in the client's favor. Although several past studies suggest that the joint provision of audit and nonaudit services may give rise to knowledge spillovers that could lead to economic rents (Palmrose 1986; Simon 1985; Simunic 1984; Turpen 1990), they do not provide direct evidence that spillovers or rents exist, and the empirical results are mixed. For example, Abdel-khalik (1990, 320) reports that he was unable to detect interdependencies between audit and nonaudit fees, a direct contrast to the findings of Simunic and Palmrose. Further, Palmrose reports a positive relation between audit fees and the nonaudit fees paid to nonincumbent firms, a finding that weakens the argument for knowledge spillovers. Thus, as Solomon (1990, 328) points out, ''. . . the impact of MAS (management advisory services) on audit pricing as well as who (i.e., the client or the auditor) benefits from knowledge spillovers (if they exist) remains an open and interesting question.'' Our empirical analysis consists of three steps. First, we demonstrate the comparability of our sample to those in prior studies by fitting prior researchers' models to our data and replicating a finding Simunic (1984) interpreted as evidence of knowledge spillovers: a positive relation between audit and nonaudit fees. Second, we use a unique data set compiled from the participating firm's internal billing records, working papers, and audit planning memos to test for a positive relation between nonaudit services and audit effort. This test is motivated by Palmrose's (1986, 410) speculation that the higher audit fees paid by clients who also purchase nonaudit services may be driven by additional audit effort. We regress audit effort on nonaudit service fees (and Palmrose's control variables), partitioning nonaudit fees into three types: tax, accounting, and other. We use three measures of audit effort: unweighted audit hours, audit hours weighted by billing rate ratios, and audit hours weighted by billing rates. We find a weakly significant, positive relation between tax services and all three audit effort measures and between accounting-related consulting services and audit hours weighted by billing-rate ratios, which suggests that additonal effort is required for audits of clients who also purchase nonaudit services. With the assumption that the demand for auditing is inelastic (Beck et al. 1988, 52-54), these results do not support the existence of audit production efficiencies from knowledge spillovers. Third, we test whether there is a positive relation between audit fees for a given level of audit effort and each of our three types of nonaudit service fees. This test is motivated by the possibility that, if the demand for auditing is elastic (Simunic 1984, 698), the observed increase in audit fees and effort could be driven by demand for more auditing by purchasers of nonaudit services (e.g., in substitution for internal control) as the result of auditors passing on cost savings from knowledge spillovers. If auditors are able to retain some of the cost savings, and thereby earn economic rents in the form of higher fees for a given level of audit effort, then our results should reveal a significant, positive relation between nonaudit and audit fees. However, we do not find a significant relation when we control for direct measures of audit effort that were not available to prior researchers. These results suggest that, although purchasers of nonaudit services pay higher audit fees than nonpurchasers, the higher fees are associated with a proportional increase in audit effort, measured in this study as unweighted and weighted audit hours. These findings are inconsistent with one interpretation of prior research: that performing nonaudit services for audit clients may provide the auditor with incentives to compromise objectivity.
