Does Selling Non-Audit Services Impair Auditor Independence? New Research Says, “Yes” Free

One of the most hotly debated audit independence issues has been the provision of non-audit services (NAS) by audit firms to their audit clients. Strong arguments have been made that such services create an economic bond between auditor and client, impairing an auditor's independence. On the other hand, equally strong opinions have been expressed that the provision of NAS increases an auditor's knowledge of the client's business, resulting in a more effective audit. Empirical studies of the potential relation between audit quality and NAS levels have found little if any evidence to support the former position, and some evidence to support the latter opinion. However, a recently published study by Causholli, Chambers, and Payne (2014) reports significant and robust evidence that NAS is associated with lower audit quality. In this article, we review this long-standing debate and summarize the newly published findings in Causholli et al. (2014).

While the issue of NAS in the U.S. may be largely a fait accompli, the question of their role on auditor independence is still an important topic of discussion for a number of reasons. First, regulatory efforts can be reversed if they are found to be ineffective in their original purpose. Second, the current U.S. regulatory prohibitions are not binding on private companies, so the independence effects of NAS are still relevant in that sector. Third, regulators outside of the U.S. are mandating NAS prohibitions or limitations (European Parliament, Council of the European Union 2014).

The debate over what is an acceptable scope of services that an audit firm should offer to audit clients has been going on for over three decades. An early discussion of the subject was Cowen (1980). Hall (1988) examines the alleged loss of independence due to the expansion of auditors' scope of practice to include non-audit services. In general, early regulatory efforts by the Securities and Exchange Commission (SEC) were greeted with skepticism. Many questioned if auditor independence was affected by the selling of NAS. There were additional concerns that restrictions on NAS could harm the economic viability of audit firms. However, others argued that the substantial sums paid by audit clients to their audit firms for NAS could create an economic bond that inevitably impairs audit firms' independence.

The spectacular accounting failures of Enron, WorldCom, and others in the early 2000s refocused this debate and resulted in two regulatory efforts that substantially changed the provision of NAS by audit firms. First, the Sarbanes-Oxley Act (SOX) prohibited the provision of most types of NAS to audit clients. Second, in 2003, the SEC issued Rule No. 33-8183 (SEC 2003) that prohibited audit partner compensation that rewards the sale of NAS to audit clients. This rule was created in response to the SEC's concerns that audit partners' compensation plans created pressure on partners to generate incremental NAS revenue from audit clients, negatively affecting their independence.

The subsequent decade saw an explosion of academic research on the question of auditor independence and the provision of NAS. Brandon and Mueller (2009) provide an excellent review of the academic research. Although an early study (Frankel, Johnson, and Nelson 2002) seemed to find evidence that the level of NAS fees were associated with lower audit quality (measured using abnormal accruals), subsequent studies found no such evidence (e.g., Ashbaugh, LaFond, and Mayhew 2003; DeFond and Francis 2005; Francis 2006; Schneider, Church, and Ely 2006; Bloomfield and Shackman 2008; Lim and Tan 2008; Habib 2012). The findings of Frankel et al. (2002) were attributed to problems in the authors' research design. Therefore, contrary to the arguments of NAS opponents and the concerns of the SEC, extensive accounting research examining the pre-SOX period, found little if any empirical evidence supporting the idea that high levels of NAS impair auditor independence. In fact, some evidence seemed to support the contrary argument that providing NAS to clients increased auditors' ability to provide a high-quality audit (e.g., Simunic 1984). Overall, the consensus of the academic studies suggested that the contemporaneous level of NAS does not, in general, have an adverse effect on audit quality. As a result, many now feel that the regulatory structures created to limit NAS have been a solution in search of a problem.

Despite the research consensus and the often expressed skepticism by auditors of the restrictions on the provision of NAS, many continued to express the belief that the provision of NAS by auditors leads to lower quality audits. Of particular concern were the compensation practices in the audit profession prior to the SEC prohibition in 2003. Prior to that time, audit firms routinely based the granting of partnership shares, and the resulting level of compensation, on the degree to which audit partners generated new NAS revenues from their clients. Art Wyatt (2003), a former senior partner at Arthur Anderson noted, “Cross-selling of a range of consulting services to audit clients became one of the most important criteria in the evaluation of audit partners. Those with the technical skills previously considered so vital to internal firm advancement found themselves with relatively less important roles.” Coffee (2006) commented that partners who successfully attracted large NAS contracts through their salesmanship abilities replaced more technically proficient audit partners who were less successful at selling NAS.

If audit partner compensation created strong incentives to bring in new NAS revenues from audit clients, then the threat to auditor independence may not have been from existing high levels of NAS, but from the potential for gaining additional NAS-related revenues in the future. Goldwasser (2002) articulated this idea when he stated, “Moreover, the threat is not the nonaudit fees that the firm has received from the client, but rather the possibility of earning such fees in the future. An accountant who hopes to obtain a large nonaudit engagement is in greater jeopardy of impaired objectivity than one who has just completed and been paid for such an engagement.” Similarly, Coffee (2006) notes his concern that the conflict of interest associated with NAS “lies not in the actual receipt of high fees, but in their expected receipt. Even the client currently paying low consulting revenues to its auditor might reverse this pattern if the auditor proved more cooperative.”

Causholli et al. (2014) set out to examine this new perspective on the question of NAS and auditor independence. All previous studies examine the association between current high levels of NAS and auditor independence, not how the potential for future NAS affects auditor independence and audit quality. Consequently, the prior studies may have been looking in the wrong place for evidence of a connection.

Causholli et al. (2014) examine audit quality in the pre-SOX years using two different measures of audit quality. First, they examine abnormal accruals—those that are higher or lower than expected. They measure abnormal accruals using a well-established model for predicting expected accruals; abnormal accruals are the difference between observed and expected accruals. If an auditor permits a client to record accruals that inappropriately either increase or decrease net income, this would be potential evidence of compromised auditor independence.

Second, they examine classification-shifting, measured by unexpected core earnings—earnings before special items. Core or pro forma earnings are very important to investors and financial analysts, and as a result there are strong incentives to manage core earnings to meet analyst forecasts. Prior accounting research has found evidence that managers who wish to report higher core earnings often do so by reclassifying core expenses into the special items section of the income statement. If material, this activity amounts to a violation of GAAP and should be reversed if discovered during a financial statement audit. In this method, core earnings are predicted using an established core earnings estimation model, and then deviations from those predictions are deemed to be unexpected core earnings. If management has systematically shifted core expenses below the line, we should observe a positive correlation between unexpected core earnings and special items.

In addition, Causholli et al. (2014) focus their attention on the subset of audit clients who have relatively low NAS fees in the current year, rather than those with high current NAS as in the prior academic research. They argue that an audit partner would be more likely to target sales efforts to clients with higher potential for new NAS. Other than clients with no interest at all in NAS, clients with low current NAS would be the most fertile ground for expanding NAS in the future.

Causholli et al. (2014) then study whether this set of clients has lower audit quality when they significantly expand NAS fees in the year subsequent to the audit. If the greatest incentive for audit partners lies in expanding NAS revenues, then the most likely place to look for impaired audit independence is in the audits of clients that subsequently purchase large amounts of NAS.

Causholli et al. (2014) first report the results of statistical regressions of abnormal accruals on variables measuring future NAS fees, current NAS fees, and a variety of control variables known from prior studies to be correlated with abnormal accruals. They find a significant association between abnormal accruals and increases in NAS fees in the subsequent year, for the set of audit clients that have low current NAS levels. In other words, audit quality suffers when audit firms successfully sell new NAS to clients most likely to be the target of selling efforts. This finding holds when abnormal accruals are limited to income-increasing or income-decreasing abnormal accruals. These results are consistent with Causholli et al.'s (2014) hypothesis that auditor independence is most likely to be impaired when auditors actively market new NAS and clients are receptive to those sales efforts.

Next, Causholli et al. (2014) test the same hypothesis by looking for positive correlations between income-decreasing special items and unexpected core earnings. Once again, they find significant evidence of classification shifting for the subset of clients most likely to be the target of NAS sales efforts and who significantly increase NAS fees in the year subsequent to the audit. These results are consistent with the hypothesis that auditor independence is most likely to be impaired when auditors successfully sell new NAS to clients who are the targets of sales efforts.

If the results from Causholli et al. (2014) are actually related to earnings management by audit clients, one would expect to see even stronger effects for the subset of clients with strong incentives to manage earnings and weaker results for the subset of clients with stronger corporate governance. They re-estimate their tests for clients with reported earnings per share just meeting or beating analyst forecasts, for clients engaged in a seasoned equity offering, and for clients with particularly strong corporate governance. As expected, they find stronger results in the first two cases (high incentives to manage earnings) and weaker results in the third case of strong corporate governance.

Given the research results provided by Causholli et al. (2014), the natural question to ask is, by what mechanism is audit quality impaired in these settings? In other words, when an audit client represents a likely source of new NAS fees, and that client is receptive to sales efforts, what occurs that results in lower audit quality?

One possible scenario is suggested by Wyatt (2003) and Coffee (2006), discussed previously. The audit profession made the cross-selling of new NAS to existing audit clients a primary metric for evaluating the performance of audit partners. Those partners who were successful in selling new NAS were rewarded with additional partnership shares and the resulting higher compensation. In addition, advancement within the firm was closely tied to successful selling efforts. In this setting, auditors faced strong incentives to sell additional NAS, particularly to audit clients who were currently not buying large amounts of these services. One can then expect that when client and auditor disagree over a financial reporting question, the auditor has strong incentives to acquiesce so as not to alienate a client who may be simultaneously considering the purchase of new NAS.

An alternate explanation for the paper's findings was suggested to us by a former audit partner who practiced during the 1990s and early 2000s. Based on this alternate explanation, the cause of the impaired independence may not have been explicit decisions by auditors to acquiesce to client demands in return for new NAS contracts. Instead, lower audit quality may have been the result of staffing decisions made by managing partners in their efforts to increase NAS. In any given year, managing partners assign engagements to audit partners based on the partners' relative skill sets and the unique characteristics of the audit client, among other reasons. Audit partners' skill sets may be classified into two groupings: those with stronger relationship-building skills (“relationship partners”) and those with stronger technical accounting and auditing skill sets (“technical partners”). When a managing partner was assigning an audit partner to a particular engagement, he or she would tend to assign a relationship partner to clients with strong potential for additional NAS fees in the future. Alternatively, clients with little or no potential for sales of new NAS would tend to be assigned to technical partners. To the degree that relationship partners possess less technical accounting and auditing skills, those clients assigned to them would receive a relatively lower quality audit. The results in Causholli et al. (2014) would be consistent with this structural, firm-level form of impaired auditor independence.

Causholli et al. (2014) also examine the same research question during the years since SOX and SEC Rule No. 33-8183 (2005–2007). They find no evidence of reduced audit quality for clients who significantly increase NAS in the year subsequent to the audit. Consequently, it appears that the regulatory intervention by Congress and the SEC has largely removed this source of impaired independence.

The larger issue of the appropriate scope of services for audit firms will continue to be a source of controversy and difference of opinion. More narrowly, the question of the appropriateness of the provision of NAS to audit clients will also continue to be intensely debated. Until recently, the body of research results on this question largely supported those who found no independence problem with the provision of NAS. The recent study by Causholli et al. (2014), however, has examined the question in a new way, and has found strong evidence that the anticipated future provision of NAS does represent a source of impaired independence in the current year.

The restrictions on the provision of most kinds of NAS contained in SOX and the prohibition of NAS selling-related compensation structures in SEC Rule No. 33-8183, look reasonable in light of these new results. The combination of these two regulatory provisions seems to have removed the potential for impaired independence. However, one may ask whether both regulatory efforts are necessary for this outcome. It is possible that the elimination of compensation incentives by the SEC may be a sufficient safeguard, and that the provisions in SOX prohibiting most forms of NAS are no longer necessary. However, we will have to wait for other academic studies to adequately answer that question.