Professional service fees and auditor objectivity.

INTRODUCTION

What causes the erosion of auditor independence? It is an intriguing question because the answer has been so elusive. The implications of audit and nonaudit services jointly provided by public accounting firms to their clients have been hotly debated for decades. Generally, the joint supply of the two services can be viewed as either leading to efficiencies (Simunic 1984) or impaired objectivity (Frankel et al. 2002). In 2000, the auditing profession narrowly avoided a Securities and Exchange Commission (SEC) proposal (Final Rule Release No. 33-7919, SEC 2000) to prohibit audit firms from providing nonaudit services to their audit clients. Recently, the SEC has also expressed concerns about the quality of earnings being reported by publicly traded companies. The sudden and stunning collapses of corporate giants such as Enron, WorldCom, and Global Crossing have, rightly or wrongly, brought these two issues together--depicting the audit firm as a collaborator in deception and helping to lead to the passage of the Sarbanes-Oxley Act of 2002 (H.R. 3763, U.S. Congress 2002), which severely restricts audit firms from providing consulting services.

A prominently cited study by Frankel et al. (2002) (hereinafter FJN) was the first academic study to report a disturbing tendency for auditors to acquiesce to client's demands when also providing nonaudit services (Kwak 2002). They found that roughly one-half of their 3,074 sample firms paid their audit firms more for consulting services than for audit services and that companies that use auditors as consultants were more likely to manage earnings. Seizing on the timeliness of their study as the Enron debacle was unfolding, the Wall Street Journal (Elstein 2001), BusinessWeek (Haddad 2002), and other popular business publications used FJN's study to link corporate financial fraud with consulting by the auditor. The media story went as follows: with large consulting fees at risk, audit firms will fail to perform their duties as an "honest watchdog," thereby clearing the way for financial fraud to go undetected (BusinessWeek 2001).

For many, the reported connection between consulting services and impaired auditor independence is intuitively appealing since it provides a convenient explanation for the recent business failures tied to financial shenanigans. As the first large-sample study of audit and nonaudit fees paid to audit firms, FJN made an important first pass at discovering the link between fee levels and audit independence (Kinney and Libby 2002, 113). The supporting evidence is weak, however, due to inconsistent results across contemporaneous studies (e.g., Antle et al. 2002; Ashbaugh et al. 2003; Chung and Kallapur 2003; Francis and Ke 2003). Given the intense debate of the effects of nonaudit services on auditor independence and its implications for the future of the profession, we argue that it is important to clear up this inconsistency.

The contribution of this paper is to "further peel the onion" to find out what is driving the FJN findings. Using data from 4,148 company proxy statements filed with the SEC in 2001, we reexamine the association between consulting services and corporation earnings management. Similar to FJN, our initial tests support a systematic relationship between earnings management and the ratio of consulting fees to total fees. Analysis conducted on the sample partitioned into quartiles based on total assets reveal that this relationship is unique to the second smallest quartile of firms. In additional tests, we find that the relationship disappears in both the full sample and the quartile samples when controlling for initial public offers (IPO), industry, and recent asset growth (Kothari et al. 2001). Specifically, we find that a disproportionately large percentage of IPO firms are in the second smallest total asset quartile. These firms were in the e-commerce, biomedical research, pharmaceutical research, or telecommunications industry and experienced one-year growth rates greater than 100 percent and as much as 3,515 percent in the extreme. Consequently, our analysis took these firms into consideration. After doing so, we find no evidence to support the contention that the amount of fees paid to audit firms affects auditor independence. Hence, notwithstanding troubling cases such as Enron, the joint provision of audit and nonaudit services does not appear to systematically cause a loss of auditor objectivity as is widely perceived by the media and others.

While we investigate the FJN discretionary accrual tests, Francis and Ke (2003) investigate the FJN earnings benchmark tests and find that the FJN earnings benchmark results are due to the inclusion of observations with large negative earnings surprises. When those observations are removed, the earnings benchmark results disappear. Our results on discretionary accruals, along with Francis and Ke's (2003) results on earnings benchmarks, provide intuitive explanations for the inconsistent results in other recent fee and independence studies.

Kinney and Libby (2002) point out a number of concerns with regard to the FJN paper that also apply to related papers such as ours. First, the data is from a single year (i.e., FY 2000), which may lead to idiosyncratic results. Since the year in question was characterized by an infatuation with e-commerce, a merger boom, and IPOs tied to the "new economy," the results seem idiosyncratic indeed. Second, classification errors between audit fees and nonaudit fees made by registrants add potential noise to the analysis. Third, the analysis depends on the strength of the cross-sectional Jones (1991) discretionary accrual computation serving as evidence of earnings manipulation and, by extension, a loss of auditor independence.