Breaking the Big Four brand’s halo effect precisely: evidence from the association between RMM coverage ratios and integrated audit effectiveness

The term “halo effect” (interchangeably halo error), coined by Edward Thorndike in the same-name book, refers to the tendency for positive impressions of a brand, product, company, or person in one area to influence one's opinion or feelings in other areas. Generally labeled as a mental heuristic and cognitive bias, the halo effect in a judgment reflects one's preferences, prejudices, ideology, aspirations, and social perception. In the context of the Big Four brand, a simple example of the halo effect is that after noticing the Big Four auditors’ large size, reputation, and exemplary performance in a single, stand-alone IC (FS) audit, one assumes that Big Four auditors can consistently deliver an integrated audit engagement, regardless of the audit effort expended or the audit approach executed.

PCAOB has released a series of Auditing Standards (AS) since 2002 to guide the integrated audits in the post-SOX (Sarbanes–Oxley Act) era, e.g., No. 4 Reporting on whether a previously reported material weakness continues to exist, No. 5 An audit of internal control over financial reporting that is integrated with an audit of financial statements; No. 8 Audit risk; No. 12 Identifying and assessing risks of material misstatement. In particular, AS 5, 8, and 12 combined mandate auditors to adopt a risk-based audit approach in the integrated audit of IC and FS for accelerated filers or large accelerated filers (collectively accelerated filers hereafter).

Oliver Hart and Bengt Holmstrom won the Nobel Prize in Economics Science in 2016 for their “contribution to contract theory.” Paul Milgrom, who won the Nobel Prize in Economics Science in 2020 “for improvements to auction theory,” also contributes to the contract theory. One key argument of the contract theory is that an executive’s compensation comprises two elements—risk allocation (interchangeably low-powered incentives) and performance-based reward (interchangeably high-powered incentives). Applying the risk allocation rule, an executive’s base salary approximately captures the quantified amount, in dollars, of the related risks that the board allocates to the executive. Simply put, the board gives an executive base salary to protect the executive from the allocated risks.

We define “audit effort” broadly. For example, total audit effort (interchangeably total audit fee) consists of the risk-prompted RMM assessments, substantive procedures, strategic risk premium, and cost-based markup.

We define integrated audit effectiveness from a third-party perspective. In particular, a year-end financial reporting package with the IC material weaknesses and FS material misstatements cleared or free of signs of executives’ earnings management indicates an effective integrated-audit engagement. Accordingly, one hidden assumption in this article is that the integrated audit effectiveness consists of the IC and FS audit effectiveness. Stated differently, an ineffective IC audit and an effective FS audit combined, or an effective IC audit and an ineffective FS audit combined, do not constitute an effective integrated-audit.

The risk allocation rule applies to the top executives exclusively from the demand side. Accordingly, a CFO’s base salary is the quantified amount, in dollars, of the financial-related risks that the board anticipates allocating to the incoming CFO, regardless of which CFO candidate is hired later. In contrast, a CFO candidate’s education, credentials, and experience are considerations from the supply side.

One hidden assumption is that CXOs’ responsibilities in a large public company are well-established and divided. To illustrate, a Chief Operation Officer (COO) performs operational-related duties primarily, a Chief Technology Officer (CTO) performs technological-related duties primarily, a CFO performs financial-related duties primarily, and a Chief Executive Officer (CEO) leads in a corporate hierarchy overseeing all functions at a higher and strategic level. In this article, we focus exclusively on the CFO role (especially the CFO’s base salary) because the CFO (and not the CEO) takes on the bulk of financial-related responsibilities (Mian 2001; Jiang et al. 2010).

Balsam et al. (2012) also find that a CFO’s base salary has zero association with CFO-specific performance. This evidence suggests that a CFO’s base salary captures the ex-ante, static financial-related risks arising from the entity’s size, complexity, and the built controls combined. On the other hand, a CFO’s activities to mitigate financial-related risks in the current accounting period are part of the CFO’s performance; thus, the compensation for the mitigating activities is typically captured by the CFO’s bonus (and not by the base salary). To provide an empirical example, Hoitash, Hoitash, and Johnstone (2012) note that the disclosure of IC material weaknesses is significantly and negatively associated with changes in a CFO’s bonus and not changes in the CFO’s base salary. This article focuses exclusively on a CFO’s base salary and does not intend to delve into the performance-based bonus literature.

One hidden assumption is that auditors will assess auditees’ RMM significantly lower immediately after auditees increase IC effectiveness over inherent risk significantly. Yes, the cooperative game theory assumes rational players.

To illustrate, on one end of the client size spectrum, Big Four auditors carefully select larger, established clients with probably more competent CFOs and corporate accountants; thus, Big Four auditors are more likely to be starting with inherently high-quality, unaudited financial reports (Johnstone and Bedard 2004). On the other end of the client size spectrum, smaller, less complex clients prefer to team up with smaller auditors, as smaller clients expect less scrutiny from smaller auditors. Accordingly, one often is uncertain whether the inherently high-quality, unaudited financial reports make Big Four auditors look stellar, or the Big Four auditors elevate the inherently high-quality, unaudited reports to an even higher-quality, audited financial reports level. Simply put, clients’ characteristics (especially the large client size), which cause them to select themselves in the Big Four’s client group, might have created the group’s high-quality, audited financial statements (Lawrence et al. 2011).

Closely related, Chaney, Jeter, and Shivakumar (2004) report that once self-selection is controlled, the audit fee premium (a proxy for earnings quality) for Big Four auditors is immediately gone among private companies.

Similarly, Nagy, Sherwood, and Zimmerman (2023) find that offices with more CPAs deliver higher quality audits.

Like comparing two companies’ net incomes, comparing two total audit fees makes little sense due to the company size noise. Alternatively, return on assets (ROA), defined as the net income over the total assets, measures how effectively a company deploys assets to generate sales and profits. With the company size noise removed, two ROAs are commensurable, especially for companies in the same industry.

To illustrate, the numerator (total audit fee) is mostly risk-prompted (e.g., pre-existing RMM, substantive procedures, strategic risk premium). In contrast, the denominator (total assets) is largely size-based.

Additionally, DeFond and Zhang (2014) emphasize that the extant literature is not conclusive on whether the audit fee premium captures higher earnings quality, monopoly pricing, or audit risk premium.

An alternative construct is the auditor industry specialization proxy. Audousset-Coulier et al. (2016) note that researchers have built 30 different proxies of auditor industry specialization (e.g., fee-based, size-based, portfolio-based, client-based, or market share-based constructs). Nevertheless, we deem the auditor industry specialization (and its 30 different proxies) a rich, separate stream of research that is out of the scope of this article.

Simply put, the pre-existing RMM on the supply side ≈ entity’s size, complexity, and the built controls combined prior to the audit ≈ ex-ante financial-related risks on the demand side (captured by the CFO’s base salary).

Given the full fee revealing equilibrium in the post-SOX era, one can reasonably expect that low-balling of audit fees is largely eliminated (e.g., Dye 1991). Similarly, the abnormally inflated audit fee due to audit inefficiency is not a concern either. For example, Francis et al. (2005) find that variance in the distribution of audit fees declines in the years following 2001, as clients who pay higher audit fees use the disclosed fee data of their closest competitors to negotiate audit fee discounts from their incumbent auditors.

We do not employ the CFO’s bonus as the denominator because the CFO’s bonus fluctuates wildly, and its lower bound could go down to zero (a zero-value denominator is undefined). Besides, we do not employ the CEO’s base salary as the denominator because the CEO’s base salary captures operational and other risks in addition to the financial-related risks. Simply put, the CEO’s base salary appears to be a noisy proxy for the pre-existing RMM.

To minimize the noise of information asymmetry that the auditor might encounter in the initial year of the audit engagement, we will exclude the company-year observations with an auditor turnover event from the sample.

For example, Lobo and Zhao (2013) argue that quarterly reports in the current accounting period, once restated, signify that the audit of the year-end financial reports will be an at-risk engagement. The reason is simple—the cumulative nature of the annual financial reports indicates that the immaterial, uncorrected errors in the quarterly reports may carry forward to the year-end and continue to build up.

AS 15 suggests that any standard audit service can deliver a low-risk audit engagement. Since an at-risk engagement poses more challenges to auditors, the at-risk engagement serves as an ideal platform to discern underperformers.

Nevertheless, we will try the RMM coverage ratio as a continuous variable in Additional Tests 7.1.

Not all PCAOB-registered auditors execute the risk-based audit approach strictly. For example, PCAOB inspection teams note that some auditors under-assess and under-address the IC risks identified, underreport the existing IC inadequacy, and, in turn, produce a low-quality IC and FS audit (Bhaskar et al. 2019). Interpreted differently, some auditors might be more used to the account-based (interchangeably size-based) audit approach, characterized by extensive testing of details (account balances, classes of transactions, events, and disclosures) and limited professional judgment. Notably, the account-based audit approach is prevalent for the single, stand-alone FS audits in the pre-SOX era; however, it is labeled deficient in the post-SOX era (Messier et al. 2021).

AS 5 also notes that while some auditors identify IC material weaknesses in the absence of an actual material misstatement, in many cases, auditors identify IC material weaknesses only when material misstatements are present.

Notably, whether an auditor executes the risk-based audit approach strictly in an integrated audit is not readily observable. Accordingly, we define an auditor as a strict auditor who identifies an IC material weakness when an FS material misstatement is present and, reversely, detects an FS material misstatement when an IC material weakness is present. This definition could be a hidden limitation.

We select the low frequency of IC material weaknesses (FS material misstatements) and zero signs of earnings manipulation to measure audit effectiveness because these three proxies appear objective and readily observable.

Noticeably, H1a (H1b) examines the association between the RMM coverage ratio deciles and IC (FS) audit-arm effectiveness, respectively. H2 examines the association between the RMM coverage ratio deciles and the aggregate, integrated audit effectiveness. Additionally, we differentiate audit effort (i.e., the RMM coverage ratio deciles) from the audit approach (whether the risk-based audit approach is strictly executed). Nevertheless, in this article, we take the risk-based audit approach strictly executed as given for a focused, inclusive analysis.

To illustrate, if a Form 8-K restatement report filed on April 29, 2013, restates a company’s financial reports between December 31, 2011 (restatement start date) and March 31, 2013 (restatement end date), the restated financial statements are the annual statements of 2011 and 2012, and quarterly statements of 2012Q1, 2012Q2, 2012Q3, and 2013Q1. We deem the reported material misstatement the interim material misstatement of 2013 because the auditor detected and reported the material misstatement on April 29, 2013. Notably, the auditor (audit effort and audit approach) in 2013 (and not 2011 or 2012) dug up the interim material misstatement. As a result, the interim material misstatement of 2013 indicates that some immaterial, uncorrected misstatements as of April 29, 2013, will carry forward and continue to accumulate towards the fiscal year-end of 2013 (assuming a December 31 FYE), making the 2013 year-end audit an at-risk engagement.

A mediation (emphasis added) analysis typically tests a causal link where one variable, A, affects a second variable, B, and the second variable, B, in turn, affects a third variable, C. For a classic mediation example, wealthy people tend to live longer, and the wealth-longevity relationship is partially explained by the mediating effect of wealthy people having access to better healthcare resources. Simply put, wealthy people can afford better healthcare resources, and the better healthcare resources, in turn, help wealthy people live longer. In contrast, a moderation analysis tests for when or under what conditions an effect occurs. For a moderation example, a student’s academic self-confidence moderates the relationship between task importance and the amount of test anxiety that the student feels. Nevertheless, task importance does not directly boost or hinder student’s academic self-confidence (Nie et al. 2011). Accordingly, we conduct a partial mediation (and not moderation) analysis in this article.

Variable definitions are listed in alphabetical order in “Appendix”.

Accordingly, the variable of IC_CLEAR captures a Big Four auditor who executes the risk-based audit approach strictly by tracking down an IC material weakness via the interim material misstatement detected in the FS audit-arm. Stated differently, we do not consider the 15 percent of auditors (mostly Big Four auditors) who underreported the IC material weaknesses and were later caught by the PCAOB inspections (Schroeder and Shepardson 2016) as strict Big Four auditors.

It is reasonably conceivable that after identifying and helping management clear the IC material weaknesses identified, strict Big Four auditors continue to conduct audit procedures to identify and help-management-correct IC ineffectiveness for the rest of the year and, in turn, implement an effective internal control system in place.

It is reasonably conceivable that after detecting and clearing the interim FS material misstatements, the strict Big Four auditors continue to conduct audit procedures to detect, adjust, and correct misstatements in the rest of the year and, in turn, achieve error-free annual financial statements.

AS 5 emphasizes that the risk factors auditors evaluate to identify significant accounts, disclosures, or relevant assertions for testing are the same in both audit-arms. Thus, we employ the same control variables in the two models.

AS 5, 8, 12 combined suggests that auditors execute the risk-based audit approach to eliminate or streamline redundant audit processes to reach maximum audit effectiveness and audit efficiency because the risk factors that the auditors evaluate in identifying significant accounts, disclosures, and relevant assertions are the same for both audit-arms. Accordingly, we differentiate the audit approach (how to expend effort) from the audit effort (how much effort to expend).

To illustrate, if the Big Four auditors of Decile 1 or under cannot conduct an integrated audit effectively, we can conclude that 20 percent (2 deciles/10 deciles) of the Big Four auditors underperform in an integrated audit.

For example, PCAOB released AS 5 on June 12, 2007, AS 8 on December 15, 2010, and AS 12 on December 15, 2010, respectively.

We also exclude from the sample the CFOs who play a dual role (e.g., the CFO and CEO simultaneously).

Some (but few) larger-sized companies [e.g., Apple, Microsoft, and Bank of America (BOA)] have an RMM coverage ratio of 20.00 or above, suggesting that their auditors charge a significant premium for strategic risks and cost-based markup and, therefore, leave a large margin for errors. In particular, a large numerator (e.g., the total audit fee of 100 million or so) and an average denominator (e.g., a CFO’s base salary of one (1) million) leads to a high RMM coverage ratio. For example, BOA’s total audit fee (the CFO’s base salary) is 75.70 (1.00) million for 2018; therefore, its RMM coverage ratio becomes 75.70. Notably, we winsorize the highest or lowest one percent of the RMM coverage ratios in each industry to mitigate the outlier effect.

For brevity, we present only the standardized results from the partial mediation analysis via the bootstrap method. Nevertheless, the unstandardized results (untabulated) from three-step logic regression are qualitatively similar to those in the main tests and the bootstrap method in Figs. 4 and 5.

Notably, the ineffective auditors (e.g., the 15 percent auditors in IC audits and 9.30–26.06% auditors in FS audits) include Big Four auditors and non-Big Four auditors. Besides, we acknowledge that while 20 percent of the Big Four auditors in our sample are less likely to conduct an integrated audit effectively, they might still conduct a single, stand-alone IC (FS) audit-arm effectively.

The upper bound of the RMM coverage ratios in Deciles 0–1 becomes 2.55 after we exclude the sixteen (16) industries with less than 100 company-year observations in the sample (e.g., ^#1 Agriculture, ^#5 Tobacco Products industry, ^#26 Defense industry). The reason for the exclusion is simple—the limited observations (sometimes only one or two unique companies) in an industry make the 16 industries less representative of the overall sample.

As discussed in Sect. 3.3, it makes little sense to compare two RMM coverage ratios close to each other (e.g., 2.31 and 2.30). Accordingly, we use “2.31 or under” as a rule of thumb rather than a law of physics. Stated differently, like a low income-to-debt ratio (three (3) or under) probably leading to a borrower’s default, a low RMM coverage ratio (2.31 or under) makes it difficult for the auditor to perform an integrated audit effectively.