Auditor Independence – Did Sarbanes-Oxley Go Too Far?

Economists Ink: A Brief Analysis of Policy and Litigation


View Falaschetti’s Profile
Dr. Dino D. Falaschetti is a special consultant for Economists Incorporated and Executive Director at the Property and Environment Research Center where his research focuses on environmental finance and law and economics. He has held endowed, tenured, and research faculty appointments in economics, finance, law, and policy, and also served on the staff of the President’s Council of Economic Advisers with responsibilities for regulation and financial services. This article builds on research he published in the American Law and Economics Review with James Brown and Michael Orlando.

Accounting scandals from the early 2000s (e.g., Enron, WorldCom) supported a popular view that letting accountants consult for audit clients compromises the quality of financial disclosures. Citing such arguments, U.S. legislators almost unanimously passed the now decade-old Sarbanes-Oxley Act of 2002 (SOX), part of which restricts accountants from producing non-audit services for audit clients. But this restriction received little support from corporate governance scholarship. Over twenty years before SOX became law, for example, the Securities and Exchange Commission’s (SEC’s)Accounting Series Release No. 250: Disclosure of Relationships with Independent Public Accountants required companies to publicly disclose fees paid to auditors for non-audit services. Researchers concluded that markets placed little value on these disclosures. Findings like these supported objections not only to the SEC disclosure mandate, which was withdrawn in 1982, but also to the SOX restriction twenty years later.

Those who objected to the disclosure mandate and the SOX restriction ignored several sources of bias in evaluating corporate governance events. First, their statistical analyses assume that event study data can show an effect in only one direction – for example, that consulting may compromise but never benefit audit relationships. At least in theory, an auditor’s consulting relationship with its client can improve the quality of financial disclosures by leveraging scope economies from jointly producing audit and consulting services. Methodologies that assume away such efficiencies may provide inaccurate results.

Second, studies of whether governance features like auditor independence create material effects can be biased if they inadequately control for investors’ anticipating information before its public disclosure. For example, when investors perfectly predict an auditor’s lack of independence, disclosures can show little if any correlation with corresponding stock prices, even if independence truly matters for earnings quality. This problem may cause an event study to incorrectly dismiss hypotheses about how governance attributes affect the quality of financial disclosures.

Third, event studies often estimate whether new information caused an abnormal response from associated security prices while ignoring other market responses. For example, if auditor independence influences the quality of information in financial disclosures, news about that independence can be associated not only with significant changes in security prices and thus investors’ returns but also with the variability of returns. As disclosures become more informative about a corporation’s fundamental value, market estimates of those values will become more precise (less variable). News can thus affect not only investors’ valuation of a corporation, but also the size of errors in forecasting future performance.

Finally, conventional event study analyses tend to ignore how governance features adopted by one firm might affect the performance of other firms. Market discipline can effectively put a price on whether a corporation governs itself well but may not accurately price how one firm’s governance affects the governance of others. Theoretical research has highlighted the potential for such governance “pollution” and its implications for policy. Nevertheless, event studies that are commonly used to evaluate policy prescriptions like those in SOX typically ignore the possibility of such third-party effects.

Taken together, these problems highlight several dimensions in which conventionally structured event studies can miss evidence for or against material effects from governance decisions and institutions. Once these difficulties are addressed, event studies and related empirical analyses indicate that auditor independence improves earnings quality, but the economic consequences of this effect may be small. This research finds little evidence of external effects from a client’s choice of auditor independence, and thus does not support the SOX proscription on corporations’ using the same firm for audit and non-audit services.

Research that takes account of the issues discussed above can shed light on many issues involving corporate governance. For example, required disclosures about executive compensation were recently expanded in the United States, and a reduction in the level of executive compensation that is deductible for tax purposes is being considered in several countries. Whether such measures can strengthen market discipline can be tested more precisely with empirical methods that view market data through firmly grounded corporate governance models.