To Hide Behind the Mask of Mandates: Disguised Opinion Shopping Under Mandatory Audit Firm Rotation and Retention in Korea

Abstract

This study investigates whether audit tenure mandates—designed to curb managerial discretion—may unintentionally enable disguised opinion shopping. Specifically, it examines whether firms benefit from observed mandates that align with their unobservable preferences, despite appearing to comply with mandatory audit firm rotation or retention rules. A counterfactual framework is developed to estimate firms’ preference for switching or retention in the absence of regulation, allowing identification of strategic alignment under constraint. Empirical analysis using Korean data from 2000 to 2009 reveals that firms classified as disguised opinion shoppers are more likely to receive unmodified audit opinions and exhibit lower audit quality, as indicated by higher discretionary accruals and more frequent reporting irregularities. These effects are concentrated under mandatory retention and not observed under rotation, suggesting that forced auditor turnover weakens firms’ ability to secure favorable outcomes. Additional evidence shows that these firms are more likely to retain the same auditor after mandates expire, consistent with a reward-for-accommodation mechanism. Thus, this study not only provides empirical evidence that opinion shopping can persist under auditor tenure mandates, but also introduces a novel method for identifying such behavior when traditional signals—such as voluntary dismissals—are unavailable. These findings inform ongoing regulatory debates on the effectiveness of tenure-based reforms.

1. Introduction

Whether firms continue to engage in opinion shopping despite regulations implemented through mandatory audit firm retention (i.e., minimum auditor tenure) and mandatory audit firm rotation (i.e., periodic changes in auditors) remains a central concern in research, as it speaks directly to the effectiveness of these rules in constraining managerial discretion (Lennox, 1998; Lu, 2006; DeFond & Zhang, 2014). Opinion shopping typically reveals itself through voluntary auditor switches or retentions undertaken to avoid adverse reporting outcomes (Lennox, 2000). However, when auditor choice is restricted by regulation, this signaling mechanism is muted. Mandatory rotation reduces the informativeness of auditor changes, making it harder to distinguish compliance from strategic behavior (Lu, 2006; Griffin & Lont, 2010; Knechel & Vanstraelen, 2007). DeFond and Zhang (2014, p. 310) suggest that “mandatory auditor rotation may exacerbate opinion shopping by allowing clients to find a more pliable auditor under the guise of switching to a more independent one.” Similarly, mandatory retention may obscure a client’s intent to stay with a more accommodating incumbent (Lu, 2006; Hunt et al., 2021). These concerns are frequently cited as potential weaknesses of tenure mandates (PCAOB, 2011), yet there is no direct evidence on whether firms engage in opportunistic auditor alignment under these constraints.

1.1. Problem Statement

This study examines “disguised opinion shopping,” a form of strategic auditor alignment that occurs when a firm’s mandated auditor outcome—retention or rotation—coincides with the choice it would have preferred absent regulation. This alignment allows firms to obscure opportunistic intent under the appearance of compliance. While not all aligned outcomes reflect opportunism, they become strategic when driven by incentives to avoid unfavorable audit opinions.

Empirically identifying disguised opinion shopping presents several challenges. First, mandatory tenure rules have only recently become widespread, limiting available data. The European Union (EU) began requiring public-interest entities to rotate auditors at least every ten years starting in 2020 (European Parliament & European Council, 2014a, 2014b). Empirical evidence is so far drawn primarily from Italy (e.g., Cameran et al., 2015; Corbella et al., 2015; Cameran et al., 2016; Horton et al., 2020), South Korea (e.g., Bae et al., 2007; Kwon et al., 2014; Mali & Lim, 2018; Lee et al., 2024; Lim, 2025), Belgium (e.g., Vanstraelen, 2000), Spain (e.g., Ruiz-Barbadillo et al., 2009), South Africa (e.g., de Ricquebourg & Maroun, 2023), Iran (e.g., Kalanjati et al., 2019), India (e.g., Dayanandan & Kuntluru, 2023), the Netherlands (e.g., de Jong et al., 2020), and Indonesia (e.g., Kalanjati et al., 2019). Second, conventional models for detecting opinion shopping—most notably the second-stage auditor switching model in Lennox (2000)—assume firms are free to replace auditors, which does not hold under regulatory mandates. Third, identifying disguised opinion shopping requires recovering firms’ counterfactual preferences—decisions they would have made if unconstrained—which are inherently unobservable.

1.2. Significance of the Study

This paper addresses these challenges by examining the Korean audit market, where mandatory auditor rotation and retention coexist with voluntary engagements. Hand-collected data from 2000 to 2009 are used to develop a two-step empirical framework for estimating firms’ counterfactual auditor preferences when tenure mandates constrain auditor choice. The likelihood of receiving a modified opinion from both incumbent and successor auditors is first estimated using voluntary engagements. These estimates are then applied to mandated firms to predict the likelihood of modification under retention and switching. Following Lennox (2000), the auditor associated with the lower predicted modification probability is classified as the firm’s preferred outcome. Firms whose mandated outcome aligns with this preference are identified as disguised opinion shoppers.

The results show that disguised opinion shoppers receive more favorable audit outcomes than non-opinion shoppers, but the effects are asymmetric. Under mandatory rotation, no benefit is observed—suggesting that the disruption and loss in client-specific knowledge from auditor turnover may outweigh opportunistic gains. Under mandatory retention, however, these firms are more likely to receive unmodified opinions, exhibit greater discretionary accruals, report more accounting irregularities, and rehire the auditor who gave favorable treatments—consistent with the notion that weak market disciplinary forces and moral hazard elevate the risk of auditor acquiescence. These findings are robust across multiple thresholds for inferring switching preferences and definitions of modified opinions. Opinion shopping is not eliminated by mandatory regulations—it is merely transformed.

This study contributes to the growing literature that questions the effectiveness of audit tenure regulation by providing the first empirical evidence of disguised opinion shopping. Prior studies document unintended consequences of mandatory retention and rotation, including adverse effects on market reactions, audit fees, and audit quality (Vanstraelen, 2000; Bae et al., 2007; Ruiz-Barbadillo et al., 2009; Chi et al., 2013; Kwon et al., 2014; Cameran et al., 2015, 2016; Corbella et al., 2015; Reid & Carcello, 2017; Horton et al., 2020). This study is the first to directly address long-standing concerns by regulators and scholars that such mandates may obscure opportunistic auditor selection (PCAOB, 2011; DeFond & Zhang, 2014).

This paper extends the opinion shopping literature to settings in which auditor choice is externally constrained. While prior studies identify opinion shopping through observable dismissals in voluntary settings (Lennox, 2000; Chen et al., 2016; Newton et al., 2016; Chung et al., 2019), this study examines a regulated environment in which such signals are suppressed. This study provides rare evidence that firms may still engage in strategic alignment under mandatory retention or rotation, even when discretion is removed.

The findings also inform the broader literature on audit quality by demonstrating that disguised opinion shopping is associated with more favorable audit opinions, higher discretionary accruals, and more frequent accounting irregularities. These firms are also more likely to reappoint the same auditor once mandates expire, consistent with a reward mechanism for prior accommodation. These results directly respond to DeFond and Zhang’s (2014) call for direct evidence on how opinion shopping affects audit quality.

Methodologically, this study contributes by introducing a counterfactual framework to infer firms’ latent auditor preferences when choice is constrained. While existing models rely on observed switching behavior, this approach enables the detection of strategic alignment even when preferences are unobservable. This framework provides a foundation for future research on hidden auditor–client dynamics in institutional environments that limit discretion. It also provides practical tools for regulators, audit committees, and institutional investors seeking to monitor opinion shopping under mandatory tenure rules.

Finally, this paper contributes to the literature on explanatory language in unqualified audit reports (e.g., Choi et al., 2023; Klevak et al., 2023; Elsayed et al., 2023; de Ricquebourg & Maroun, 2023; Wang et al., 2024; Chiu et al., 2025). While audit opinions may remain unqualified, the audit report—auditors’ primary communication channel with financial statement users—can still convey underlying concerns through variations in explanatory language (Choi et al., 2023; de Ricquebourg & Maroun, 2023; Wang et al., 2024; Chiu et al., 2025). Klevak et al. (2023) show that more extensive explanatory disclosures are associated with increased stock price volatility and forecast dispersion, suggesting that investors interpret them as signals of heightened uncertainty. Elsayed et al. (2023) find that such disclosures affect firms’ risk fundamentals and market responses, including cost of equity and liquidity.

Extending this literature, this study leverages the Korean audit environment, where explanatory language is not only prevalent (Lee et al., 2024) but encouraged by regulators (FSS, 1997–2009; Korea Economic Daily, 1996). Korean regulators promote the use of explanatory language to flag red-flag issues such as frequent financing, ownership changes, and related-party transactions—risks that are especially salient in Korea’s complex business group structures (Chang & Hong, 2000). This study provides new evidence on how tenure mandates influence firms’ incentives to strategically align with auditors to avoid such disclosures.

2. Background and Prior Literature

2.1. Mandatory Audit Firm Rotation and Retention in Korea

Korea’s audit environment provides a unique institutional setting in which both mandatory retention and rotation policies have shaped auditor tenure. For KOSPI-listed firms, mandatory retention was introduced in 1997 and became fully enforced in 2000, requiring firms to retain the same auditor for three consecutive years, with selection discretion granted only in the first year of each engagement cycle. KOSDAQ-listed firms became subject to the same rule in 2002.

In 2006, Korea adopted a mandatory rotation policy prohibiting firms from reappointing the same audit firm beyond two consecutive terms. This rule applied retroactively, compelling firms with six or more years of continuous engagement to rotate in 2006. The policy was met with immediate resistance: the Financial Supervisory Service (FSS) proposed a repeal in 2008 (FSS, 2008), and the Korean Congress passed an amendment in 2009 to eliminate the rotation requirement, effective from 2011.

Although both policies restrict auditor choice, they operate in opposite directions and triggered different institutional responses. Mandatory rotation shortened auditor tenure and faced strong resistance from audit firms, along with increased scrutiny from regulators and investors (Herald Business Newspaper, 2002). In contrast, mandatory retention extended auditor tenure and encountered little resistance, likely because it reinforced existing client–auditor relationships.

2.2. Opinion Shopping Literature

2.2.1. Opinion Shopping Under Voluntary Settings

Recent studies broaden the understanding of opinion shopping by demonstrating its use not only to avert unfavorable reporting outcomes but also to enable opportunistic disclosure. Auditor changes have been linked to the avoidance of going-concern opinions (Chung et al., 2019), adverse internal control assessments (Newton et al., 2016; Amin et al., 2021), modified audit opinions (Chen et al., 2016), and the underreporting of impairments and misstatements (Ayres et al., 2019; Hunt et al., 2021; Singer & Zhang, 2022). DeFond et al. (2025) further show that managers may dismiss incumbent auditors to pursue income-increasing accounting estimates.

The consequences of such strategically motivated dismissals are often adverse. DeFond et al. (2025) report that these dismissals are followed by higher rates of earnings restatements, SEC comment letters, and earnings target attainment, suggesting that successor auditors may enable more aggressive reporting. Singer and Zhang (2022) find that opinion shopping coincides with deteriorated financial reporting and increased turnover among executives and audit committee members. Similarly, Chung et al. (2019) show that distressed firms replacing auditors to avoid going-concern opinions are more likely to incur Type II audit errors and subsequent restatements, while those retaining their auditors do not exhibit comparable declines in audit quality.

The likelihood and success of opinion shopping depend on corporate governance mechanisms and institutional audit settings. Amin et al. (2021) find the practice more common during periods of elevated investor sentiment, particularly among firms with low institutional ownership that frequently switch from Big 4 to non-Big 4 auditors. Du et al. (2023) identify dialectal connectedness between CEOs and auditors as a relational channel that compromises auditor independence, though this effect weakens for Big 10 auditors. Xing et al. (2023) show that firms are more likely to replace auditors following goodwill impairment opinions when internal controls are weak or analyst scrutiny is low. Newton et al. (2016) find that clients in competitive audit markets, especially those dismissing auditors late in the reporting cycle, are more likely to obtain clean internal control opinions—highlighting how market structure and timing facilitate successful opinion shopping.

However, not all auditor changes reflect opportunism. Chen et al. (2019) find that firms switching audit offices within the same network improve earnings quality and reduce bankruptcy risk, suggesting that some dismissals reflect informational needs rather than efforts to influence audit outcomes.

Together, these findings underscore that opinion shopping under voluntary settings is both prevalent and strategically motivated, shaped by firm incentives, market dynamics, and auditor–client relationships.

2.2.2. Opinion Shopping Under Tenure Mandates

While prior studies have examined the capital market consequences of opinion shopping in Korea, they generally assume firms have discretion over auditor choice. Chung et al. (2021) find that auditor switches associated with opinion shopping are followed by weaker earnings response coefficients. Chung and Lee (2024) link such behavior to aggressive tax avoidance, particularly when firms move to non-Big 4 auditors. Chung et al. (2025) report that suspected opinion shoppers face higher borrowing costs, especially under weak governance conditions.

These studies, however, primarily use qualified, adverse, or disclaimer audit opinions as proxies for unfavorable outcomes—a limitation in the Korean context, where such opinions are exceedingly rare (Choi et al., 2023). In contrast, recent research shows that unqualified opinions still vary in their informational value. Although uniformly labeled as unqualified, audit reports differ meaningfully in the content and tone of explanatory language. de Ricquebourg and Maroun (2023) show that shifts in KAMs can signal subtle changes in auditor concern. Chiu et al. (2025) demonstrate that audit partner rotation within the same firm often results in changes to KAM topics and procedures, reflecting partner-specific discretion. Wang et al. (2024) find that auditors tailor KAMs to financially risky clients, though this differentiation weakens under client pressure. In the Korean setting, Choi et al. (2023) show that negatively toned explanatory language is associated with higher audit fees, greater perceived risk, and lower accounting quality—suggesting that narrative disclosures convey meaningful signals even when opinions remain unqualified.

Moreover, prior studies often overlook the institutional constraints imposed by Korea’s mandatory retention, which limits firms’ ability to switch auditors in response to unfavorable audit outcomes. A notable exception is Lee et al. (2024), who broaden audit quality measures by analyzing explanatory language within unqualified opinions and incorporating both retention and rotation requirements into their empirical design.

This study builds on Lee et al. (2024) and extends prior literature by incorporating detailed audit opinion data including explanatory language and explicitly distinguishing engagements as mandatory or voluntary—recognizing that institutional constraints affect auditor behavior and client incentives. It embeds these distinctions into an audit opinion model to isolate voluntary engagements and estimate the audit outcomes firms would receive if they retained discretion in auditor selection.

2.3. Audit Quality Literature

2.3.1. Audit Quality Under Voluntary Settings

Audit quality is commonly proxied by discretionary accruals (Kothari, 2001; Myers et al., 2003). High accruals indicate earnings management and low audit quality, while low accruals reflect conservatism and strong oversight (DeFond & Subramanyam, 1998; Bartov et al., 2000).

Although findings remain mixed (e.g., Alhazmi et al., 2024), longer auditor tenure is generally associated with higher audit quality. Studies find that extended tenure reduces accruals and increases the likelihood of modified opinions before failures (Myers et al., 2003; Geiger & Raghunandan, 2002; Carcello & Nagy, 2004; Knechel & Vanstraelen, 2007; DeFond & Zhang, 2014).

Another line of research adopts a nonlinear perspective, showing that audit quality initially improves as auditors gain familiarity, but may decline in later years due to reduced independence (Johnson et al., 2002; Davis et al., 2009; Boone et al., 2008; Jenkins & Velury, 2008; Mustikarini & Adhariani, 2022).

Together, the findings suggest that while audit quality often improves with tenure, excessive familiarity may eventually reduce auditor independence.

2.3.2. Audit Quality Under Tenure Mandates

Cross-Jurisdictional Evidence on Mandatory Tenure

Cross-country research presents mixed evidence on mandatory retention and rotation. In Belgium, Vanstraelen (2000) finds that modified opinions increase in the final year of the three-year mandatory retention period. In Spain, Ruiz-Barbadillo et al. (2009) observed no change in going-concern reporting after a proposed mandatory rotation policy was reversed, indicating minimal regulatory impact.

Findings from Italy are mixed. Cameran et al. (2015) report that successor auditors spend more hours, charge lower fees, and tolerate higher abnormal accruals—consistent with lowballing and weaker audit quality. In contrast, Corbella et al. (2015) find that, controlling for fees, rotation is associated with lower abnormal accruals. Cameran et al. (2016) show that audit conservatism increases in the final years of tenure but declines in early years due to steep learning curves. Horton et al. (2020) report that mandatory firm rotation yields no incremental audit quality improvements—measured by abnormal accruals and earnings response coefficients—beyond partner rotation.

In Indonesia, Kalanjati et al. (2019) link repeated mandatory rotations to higher absolute discretionary accruals, driven by auditor transition frictions. In Iran, Salehi et al. (2022) find that rotation increases accrual-based earnings management, with no effect on real earnings management or audit fees—highlighting ineffectiveness under economic stress. Similar early-tenure declines in quality appear in the Netherlands and India. de Jong et al. (2020) and Dayanandan and Kuntluru (2023) both document reduced audit quality early in tenure due to auditor learning curves.

Empirical Findings from Korea’s Auditor Rotation and Retention

Empirical findings on Korea’s auditor tenure regulations are mixed. Bae et al. (2007) examine the earlier mandatory retention regime and find that firms report fewer abnormal accruals and receive more modified opinions, suggesting improved audit quality through enforced continuity. Under the later mandatory rotation regime, Kwon et al. (2014) find no significant improvements in audit quality—measured by discretionary accruals—relative to long-tenured or voluntarily rotated engagements. Although audit fees rose post-rotation, the absence of quality gains points to reduced audit efficiency. Using accrual quality, Mali and Lim (2018) similarly report no improvement. On the other hand, Kim et al. (2019) document a decline in firms’ cost of equity capital, indicating improved perceptions of auditor independence rather than changes in actual audit quality. However, Lim (2025) finds that Tobin’s q declined during the enforcement period and rebounded after repeal, suggesting investors viewed mandatory rotation as value-destroying.

Most prior studies related to mandated regulations examine these rules using regime-level comparisons. For example, Lee et al. (2024) report that mandatory retention reduces opinion shopping, while rotation increases it, relative to voluntary regimes.

This study shifts the analytical focus from aggregate regime effects to firm-level behavior under constraint. Specifically, it investigates how the alignment between a firm’s preference and its mandated auditor outcome affects audit quality. While data from voluntary regimes are used to estimate counterfactual preferences, the main inferences pertain to firms operating under tenure mandates.

3. Hypothesis Development

Auditor switching is commonly interpreted through the lens of opinion shopping, wherein firms seek more favorable audit outcomes by changing auditors. Lennox (2000) formalizes this intuition, showing that firms are more likely to switch (retain) when they expect a lower (higher) likelihood of receiving a modified opinion from a successor auditor relative to the incumbent. This framework has shaped a broad literature on voluntary auditor changes (e.g., Chen et al., 2016; Chung et al., 2019; Berglund et al., 2022; Osma et al., 2022; Zhang et al., 2022), offering evidence that auditor selection is often strategic.

This interpretation is grounded in agency theory, which conceptualizes external audits as a key mechanism for mitigating conflicts in the principal–agent relationship (Chow & Rice, 1982; Habib et al., 2019a). In this framework, shareholders (principals) delegate decision-making authority to managers (agents), whose incentives may diverge from shareholder interests due to information asymmetries. Auditors function as delegated monitors, responsible for enhancing the credibility of financial statements by offering assurance on reporting accuracy, going-concern risks, internal controls, and other key disclosures (Watts & Zimmerman, 1983; Kraakman, 1986).

Agency theory further acknowledges that auditors are themselves economic agents whose incentives may not be fully aligned with those of shareholders (Antle, 1982). As such, the monitoring role of the auditor is susceptible to erosion. Long-standing client relationships, fee dependence, and client pressures can impair auditor objectivity and independence, leading to acquiescence to managerial preferences. These agency problems may manifest as a moral hazard—where auditors reduce effort or professional skepticism—or adverse selection—where less competent or biased auditors are retained or chosen (Chow & Rice, 1982; Habib et al., 2019a, 2019b; Velte & Loy, 2018).

Disguised opinion shopping exemplifies these agency problems. Even under mandatory retention or rotation rules designed to constrain auditor choice, firms may still achieve alignment with auditors predisposed to issue favorable opinions within the constraints of tenure mandates. Such alignment, though formally compliant, can undermine audit quality by weakening the auditor’s gatekeeping function and sustaining information asymmetries (Kraakman, 1986).

To detect such behavior, a counterfactual framework is developed that estimates firms’ preferred auditor outcome—retention or switching—in the absence of regulatory constraints. Using voluntary data, the likelihood of a modified opinion under both choices is predicted and compared to the mandated outcome. This produces a two-by-two framework: firms are categorized by whether regulation requires them to retain or switch, and whether their inferred preference is to retain or switch.

Firms are considered “aligned” when their mandated outcome matches their predicted preference—either required to switch and preferring to switch (Quadrant I), or required to retain and preferring to retain (Quadrant III). They are “misaligned” when the mandate conflicts with their predicted preference—either required to switch but preferring to retain (Quadrant II), or required to retain but preferring to switch (Quadrant IV). See Figure 1 for an illustration of this classification.

This asymmetric structure echoes findings from prior research on mandatory rotation. Lu and Sivaramakrishnan (2009) show that forced auditor turnover can generate both beneficial and disruptive effects on investment efficiency. Deng et al. (2019) similarly find that rotation mandates allow some firms to exit low-quality auditor relationships, but force others to abandon high-quality ones. These studies suggest that uniform tenure mandates may yield heterogeneous firm-level outcomes.

Building on this insight, alignment under constraint creates the possibility of strategic auditor selection masked by regulatory compliance. In Quadrant I, a mandate facilitates a preferred switch, potentially enabling firms to replace a stricter auditor. In Quadrant III, retention allows firms to continue relationships they already favor, potentially preserving accommodative ties. In both cases, firms achieve preferred outcomes while formally complying with tenure rules.

This interpretation is consistent with prior research linking opinion shopping to diminished audit quality. Strategic auditor selection has been associated with avoiding going-concern opinions (Chung et al., 2019), adverse ICFR assessments (Newton et al., 2016; Amin et al., 2021), modified audit opinions (Chen et al., 2016), and underreporting of impairments or misstatements (Ayres et al., 2019; Singer & Zhang, 2022; Hunt et al., 2021). DeFond et al. (2025) further show that opinion shopping is correlated with income-increasing estimates, restatements, and the likelihood of meeting or beating earnings benchmarks.

If alignment between mandated outcomes and firm preferences reflects strategically motivated auditor selection cloaked in compliance, then aligned firms are expected to receive more favorable audit treatment than misaligned firms. While alignment does not guarantee accommodation, it increases the likelihood that firms achieve outcomes consistent with their underlying incentives.

To test this, this study examines audit outcomes along four dimensions: (1) issuance of a modified opinion, (2) magnitude of discretionary accruals, (3) incidence of financial reporting irregularities measured by financial statement divergence score (Amiram et al., 2015), and (4) likelihood of auditor reappointment. Favorable outcomes are defined as receiving an unmodified opinion, reporting higher discretionary accruals and irregularities, and continuing the relationship with the same auditor. This leads to the following testable hypothesis:

H1: 

Firms whose mandated auditor outcome aligns with their counterfactual preference are more likely to receive favorable audit outcomes than firms with misaligned outcomes.

Mandatory rotation disrupts auditor–client continuity and firm-specific learning, which can reduce audit quality (Florio, 2024). New auditors face steep learning curves, familiarizing themselves with the client’s operations, internal controls, and reporting processes while raising the risk of oversight—particularly in areas demanding professional judgment (Geiger & Raghunandan, 2002; Carcello & Nagy, 2004; Cameran et al., 2015; Casterella & Johnston, 2013; Catanach & Walker, 1999). Early-tenure audits are also associated with increased costs and reduced efficiency (Myers et al., 2003; Jackson et al., 2008). Lim (2025) emphasizes that successor auditors may overlook unusual transactions or major shifts in the client’s environment due to limited client-specific knowledge during initial years. These transitional frictions heighten the risk of Type II audit failure, in which auditors issue clean opinions despite conditions that warrant a going-concern modification (Chung et al., 2019; Velte, 2023). Supporting this concern, Cowle et al. (2023) find that firms engaging in frequent auditor switching exhibit lower audit quality—evidenced by higher misstatement risk and more aggressive accruals—relative to firms that switch auditors only once. Their findings suggest that repeated disruptions to auditor continuity can compound learning-related frictions and further degrade audit effectiveness.

Empirical evidence from mandated settings supports these concerns. Forced rotations have been linked to extended reporting delays (Tanyi et al., 2010), weakened audit committee oversight (Fontaine et al., 2016), and diminished audit effectiveness due to transition-related disruptions (Harber et al., 2020). In Korea, Kwon et al. (2014) document a decline in audit quality following the implementation of rotation requirements. Furthermore, mandatory limits on tenure may undermine auditors’ incentives to invest in client-specific knowledge (Lennox, 2014), and auditor transitions are frequently associated with heightened regulatory scrutiny (Firth et al., 2011; Chen et al., 2019).

Mandatory rotation also weakens the channels through which opinion shopping typically operates. Economically, fixed engagement terms reduce the scope for quasi-rents, limiting auditors’ incentives to accommodate clients in exchange for long-term fee continuity (Zhang, 1999). When auditors enter engagements with a predetermined end date, the marginal value of client retention is diminished. Reputational pressures are likewise attenuated: when auditor turnover is mandated, disengagement no longer signals the audit partner’s performance failure (Ruiz-Barbadillo et al., 2006).

Given that all rotated firms experience these transitional disruptions and reduced incentives to compromise, outcomes may not differ based on preference alignment. In particular, if competence-related frictions are binding, no systematic difference in audit outcomes should emerge between aligned and misaligned firms.

H2 (Auditor learning/Competence effects): 

Among firms subject to mandatory rotation, audit outcomes do not differ between those whose preferences align with the mandate and those whose preferences do not.

Extended auditor–client relationships may gradually impair auditor independence. Familiarity formed through repeated engagements can reduce professional skepticism and lead auditors to over-rely on management representations or past assessments (Casterella & Johnston, 2013; Daniels & Booker, 2011; Lennox et al., 2014). Carey and Simnett (2006) find that longer tenure is associated with a decreased likelihood of issuing qualified audit opinions, consistent with a reduction in objectivity. These dynamics are often compounded by economic incentives. Auditors deriving quasi-rents from client-specific investments may be reluctant to issue unfavorable opinions that jeopardize future engagements (DeAngelo, 1981; Catanach & Walker, 1999). Over time, audit procedures may also become routinized, diminishing responsiveness to evolving risk profiles.

Mandatory retention may exacerbate these concerns, when scrutiny is low. Although some studies associate longer auditor tenure with enhanced earnings quality (Geiger & Raghunandan, 2002; Johnson et al., 2002; Myers et al., 2003; Carcello & Nagy, 2004; Jadiyappa et al., 2021; Kamarudin et al., 2022), these results generally arise in voluntary audit markets where retention reflects mutual willingness to continue the engagement. In such contexts, auditors retain the discretion to withdraw from problematic clients, and the act of remaining may signal confidence in the client’s reporting or a commitment to enforce discipline (Chung et al., 2019). By contrast, mandatory retention regimes eliminate such discretion. Under Korea’s three-year retention rule, auditors lacked procedural discretion to disengage, even when audit concerns emerged during the engagement. Until 2011, resignation due to undue influence or unreasonable client demands was not formally permitted.

This structural rigidity has important implications when clients are motivated to obtain more favorable audit outcomes. Under a mandatory retention regime, once the auditor is appointed, the client faces limited opportunities to initiate a change. Firms flagged for disguised opinion shopping, in particular, may attempt to retain auditors perceived as more pliable in order to avoid unfavorable disclosures (Lee et al., 2024). The auditor, in turn, operates with reduced exposure to market discipline during the fixed engagement period. In such cases, the cost of compromising independence is low—regulatory enforcement tends to be delayed or diffuse (Zhang et al., 2022; Yao & Xue, 2019; Quick et al., 2024), and the disciplining threat of auditor dismissal is absent. Auditors may rationally accommodate client preferences in these settings, especially when such accommodation preserves relational continuity or positions the auditor favorably for future engagements.

Mandatory retention also introduces a moral hazard problem (Chow & Rice, 1982; Habib et al., 2019a, 2019b; Velte & Loy, 2018): the policy intends to promote independence by removing fee-based competition and client pressure, but it can also insulate auditors from accountability. When short-term costs of compromise are negligible and engagement continuity is guaranteed, auditors may exert less effort or adopt a more lenient posture toward client demands (Dye, 1993; Watts & Zimmerman, 1983). Prior research highlights that the threat of dismissal plays a critical role in disciplining audit behavior (DeFond & Francis, 2005; Carcello & Neal, 2003). By removing this mechanism, mandatory retention may reduce both perceived and actual independence—particularly among engagements where clients have strong incentives to avoid scrutiny.

In sum, audit quality under mandatory retention is not uniform. Retained clients flagged for disguised opinion shopping are more likely to exert pressure to avoid unfavorable outcomes, and auditors operating under restricted exit options and reduced market discipline may be more likely to yield. These conditions jointly increase the risk of compromised independence and lower audit quality.

H3 (Auditor closeness/Independence effects): 

Among firms mandated to retain their auditor, those whose counterfactual preference also favors retention are more likely to receive favorable audit outcomes than those who would have preferred to switch auditors.

4. Research Methodology

4.1. Sample and Data

The sample consists of firms listed on the KOSPI and KOSDAQ exchanges between 2000 and 2009. Financial and accounting data are obtained from the TS2000 database, compiled by Korea Companies Information (KOCOinfo). Audit reports are hand-collected from the Data Analysis, Retrieval and Transfer System (DART), the electronic disclosure platform maintained by Korea’s Financial Supervisory Service. Audit opinions, auditor signature dates, audit fees, and audit hours are extracted from these filings.

Table 1. Sample selection procedures.

Sample Selection Criteria	Number of Firm-Years
Number of KOSPI- and KOSDAQ-listed firms	20,843
Less: non-December fiscal year-end firms	1646
Less: firms in non-compliance with mandatory rotation or retention	755
Less: firms in the financial and insurance industries	350
Less: firms that changed their fiscal year-end	45
Less: firms missing relevant data	7262
Both mandatory and voluntary sample for models (1) and (2)	10,785
Less: voluntary auditor switching and retention	3329
Mandatory sample for models (3) and (4)	7456
Less: firms missing relevant data	6422
Post-mandate sample for model (5)	1034

summarizes the sample selection process. Starting from 20,843 firm-year observations, the following are excluded: firms with non-December fiscal year-ends, firms in the financial and insurance industries, firms that changed the fiscal year-end during the sample window, observations with missing key data, and firms not in compliance with Korea’s mandatory auditor tenure requirements.

The final sample includes auditor engagements classified as mandatory rotation, mandatory retention, or voluntary appointment. Models (1) and (2) include both mandated and voluntary observations to estimate counterfactual auditor preferences. Models (3) and (4) restrict the analysis to engagements subject to tenure mandates. Model (5) examines auditor retention behavior following the expiration of such mandates.

4.2. Data Collection and Coding Procedures

Audit opinions are first classified into four standard types: unqualified, qualified, adverse, and disclaimer. Unqualified opinions are further classified based on the presence and nature of explanatory language. These are categorized into four subgroups: unqualified with no explanatory language, and unqualified with harmless, potentially negative, or harmful explanatory language.

The classification of explanatory language proceeds in three stages. First, following Lee et al. (2024), a sample of 500 unmodified audit reports is manually reviewed to identify keywords and phrase structures corresponding to 20 distinct types of explanatory language. Second, the full set of unqualified audit reports is classified using text-parsing software that applies the keyword-based rules derived from the manual review. Third, classification accuracy is validated by manually reviewing a random sample of 300 audit reports, with over 97 percent of reports correctly assigned.

The 20 types of explanatory language are grouped into three signaling categories, based on their informational severity and potential for investor concern (Lennox, 2002; Czerney et al., 2014, 2019; Chen et al., 2016; Lee et al., 2024). Harmless language includes routine disclosures such as reliance on other auditors, minor comparability adjustments, and changes in accounting principles. Potentially negative language includes matters flagged by the Financial Supervisory Service (FSS, 2014) as cautionary signals—such as frequent capital raising, unusual equity transactions, or internal restructuring within Chaebol. Harmful language includes going-concern uncertainties, financial distress, restatements, reorganization, legal risks, and related-party transactions. Please see the notes in

Table 2. Audit opinion.

	(1) Voluntary Auditor Switching and Retention		(2) Mandatory Auditor Switching and Retention		(3) Total
Opinions	# of Obs.	%	# of Obs.	%	# of Obs.	%
Unqualified opinions with no explanatory language	696	20.68	2058	27.6	2754	25.54
Unqualified opinions with      harmless explanatory language	304	9.11	688	9.23	992	9.2
Unqualified opinions with      potentially negative explanatory language	405	12.14	663	8.89	1068	9.9
Unqualified opinions with      harmful explanatory language	1872	56.12	3919	52.56	5791	53.69
Qualified opinions	38	1.14	71	0.95	102	0.95
Adverse opinions	1	0.03	2	0.03	3	0.03
Opinion disclaimer	20	0.6	55	0.74	75	0.7
Total	3336	100	7456	100	10,785	100

Notes: Table 2 reports the classification of audit opinions for all firms in either the voluntary or mandatory auditor sample. Following the framework proposed by Lee et al. (2024), unqualified audit opinions are categorized based on the nature of the accompanying explanatory language into three groups: harmless, potentially negative, and harmful. Harmless language includes reliance on another auditor’s report; minor comparability changes, such as name adjustments; accounting principle changes consistent with applicable standards; reclassifications of prior-year financial statements for consistency; and disclosures related to mergers, acquisitions, or divestitures. Potentially negative language suggests elevated risk but not immediate concern, and includes frequent financing activities; changes in management or controlling ownership; internal restructuring within Chaebol; unusual equity transactions; sales of significant assets; subsequent event disclosures; emphasis on business or economic risk factors; and changes in accounting estimates. Harmful language reflects more serious concerns, including going-concern uncertainties, financial distress, corporate reorganization, legal or regulatory violations, related-party transaction risks, and restatements of prior financial statements.

for a full description of the 20 types.

Table 2 compares audit opinion distributions across voluntary and mandated auditor appointments. Firms subject to mandates are significantly more likely to receive unqualified opinions with no explanatory language, while those under voluntary regimes more frequently receive unqualified opinions with negative or harmful explanatory language. Qualified, adverse, and disclaimer opinions collectively account for only 1.68% of the sample. Consistent with prior literature (e.g., Lennox, 2002; Czerney et al., 2014, 2019; Chen et al., 2016; Lee et al., 2024), modified opinions are defined to include unqualified opinions with harmful explanatory language, as well as qualified, adverse, and disclaimer opinions—comprising approximately 55% of audit reports in the sample.

4.3. Justification of Sample Period

The 2000–2009 sample period captures a regulatory environment in which both mandatory and voluntary auditor switching and retention occurred. This institutional setting introduces variation in auditor choice discretion, allowing for the identification of counterfactual auditor preferences and the analysis of auditor–client alignment under regulatory constraint.

Audit opinion variables became publicly available for all listed firms in 1999, the first year of mandatory electronic reporting through Korea’s DART system. However, because the analyses in this study rely on lagged audit opinion data, the effective sample begins in 2000—the first year in which both current and prior-year audit opinions are reliably available across the full population of listed firms.

The sample period concludes in 2009, when Korea’s National Assembly passed legislation to repeal the mandatory audit firm rotation rule, with implementation scheduled for 2011. Ending the sample prior to this policy shift minimizes the risk of capturing anticipatory changes in auditor or client behavior and ensures a consistent regulatory environment throughout the analysis. This cutoff also aligns with prior research on Korea’s audit reforms (e.g., Kwon et al., 2014). Limiting the sample to the pre-2010 period further reduces the risk of confounding effects related to changes in financial reporting standards, given that voluntary adoption of International Financial Reporting Standards (IFRS) was permitted beginning in 2009, although mandatory implementation did not occur until 2011.

4.4. Estimating Counterfactual Auditor Switching Preferences Under Tenure Mandates

To recover firms’ unobserved preferences under tenure mandates, this study extends Lennox’s (2000) first-stage audit opinion model, where the probability of receiving a modified opinion is modeled as a function of financial attributes and their interactions with auditor switching (S):

MO_i,t = β₁ (Financial Variables_i,t) + β₁S_i,t × (Financial Variables_i,t)

(1)

Specifically, to account for regulatory intervention, I introduce an indicator for mandatory tenure constraints (MAND) and its interactions:

$$\begin{array}{l}{\mathrm{MO}}_{\mathrm{i},\mathrm{t}} = {\mathsf{\beta }}_1(\mathrm{Financial} {\mathrm{Variables}}_{\mathrm{i},\mathrm{t}}) +{ \mathsf{\beta }}_2{\mathrm{S}}_{\mathrm{i},\mathrm{t}} \times (\mathrm{Financial} {\mathrm{Variables}}_{\mathrm{i},\mathrm{t}})\\ + {\mathsf{\beta }}_3{\mathrm{MAND}}_{\mathrm{i},\mathrm{t}} \times (\mathrm{Financial} {\mathrm{Variables}}_{\mathrm{i},\mathrm{t}}) + {\mathsf{\beta }}_4{\mathrm{MAND}}_{\mathrm{i},\mathrm{t}} \times {\mathrm{S}}_{\mathrm{i},\mathrm{t}} \times (\mathrm{Financial} {\mathrm{Variables}}_{\mathrm{i},\mathrm{t}})\end{array}$$

(2)

Financial covariates include audit opinion modification in the prior year (MOLAG), firm size (SIZE), leverage (LEV), liquidity (LIQ), profitability (PROF), operating cash flows (OCF), net loss (LOSS), stock return (RETURN), return volatility (VOLATILITY), book-to-market ratio (BM), capital issuance (ISSUANCE), and Big N auditor status (BIGN). Appendix A provides detailed definitions.

Using the estimated model, I calculate predicted probabilities of receiving a modified opinion under two hypothetical voluntary scenarios: (1) voluntary retention (i.e., $\mathrm{Pr}({\mathrm{M}}_{\mathrm{it}}^{\mathrm{S}=0, \mathrm{MAND}=0}=1))$ and (2) voluntary switching (i.e., $\mathrm{Pr}({\mathrm{M}}_{\mathrm{it}}^{\mathrm{S}=1, \mathrm{MAND}=0}=1))$. The firm’s inferred preference corresponds to the option with the lower estimated likelihood of modification.

Table 3. Audit reporting model by auditor tenure constraints: mandatory vs. voluntary.

Dep. =	MO
	(1)		(2)
Variable	Coeff	Z-Stat	Coeff	Z-Stat
MOLAG	2.034 ***	49.35	1.904 ***	26.88
SIZE	0.099 ***	5.71	0.091 ***	3.10
LEV	0.484 ***	4.06	0.598 ***	2.73
LIQ	0.005	0.69	0.012	0.94
LOSS	0.062	1.20	0.057	0.63
BM	13.452 **	2.37	31.440 ***	2.64
ISSUANCE	−0.022	−0.61	−0.006	−0.09
OCF	−0.322 **	−2.03	−0.228	−0.80
BIGN	−0.628 ***	−16.69	−0.433 ***	−6.46
RETURN	0.009	0.43	0.037	1.14
VOLATILITY	0.121	1.11	−0.027	−0.16
PROF	−0.988 ***	−7.59	−0.865 ***	−3.68
S	0.526	0.64	1.450	1.44
S × MOLAG	−0.719 ***	−7.81	−0.470 ***	−3.59
S × SIZE	−0.009	−0.22	−0.047	−0.88
S × LEV	0.096	0.35	−0.096	−0.27
S × LIQ	−0.020	−1.07	−0.024	−0.97
S × LOSS	−0.025	−0.21	0.119	0.70
S × BM	−7.273	−0.75	−22.760	−1.53
S × ISSUANCE	−0.073	−0.80	−0.043	−0.34
S × OCF	−0.892 **	−2.50	−1.399 ***	−2.84
S × BIGN	−0.139	−1.46	−0.085	−0.64
S × RETURN	0.077 **	2.09	0.036	0.81
S × VOLATILITY	0.289	1.24	0.061	0.33
S × PROF	0.604 **	2.31	0.995 ***	3.05
MAND			−0.034	−0.05
MAND × MOLAG			0.179 **	2.30
MAND × SIZE			0.010	0.30
MAND × LEV			−0.148	−0.57
MAND × LIQ			−0.008	−0.49
MAND × LOSS			0.003	0.03
MAND × BM			−25.183 **	−2.01
MAND × ISSUANCE			−0.030	−0.38
MAND × OCF			−0.140	−0.41
MAND × BIGN			−0.270 ***	−3.36
MAND × RETURN			−0.038	−1.04
MAND × VOLATILITY			0.250	1.11
MAND × PROF			−0.144	−0.50
S × MAND			−3.506 **	−2.01
S × MAND × MOLAG			−0.506 **	−2.56
S × MAND × SIZE			0.171 *	1.90
S × MAND × LEV			0.493	0.77
S × MAND × LIQ			0.003	0.06
S × MAND × LOSS			−0.247	−0.93
S × MAND × BM			−12.902	−0.37
S × MAND × ISSUANCE			−0.152	−0.76
S × MAND × OCF			1.443 *	1.80
S × MAND × BIGN			−0.600 ***	−2.70
S × MAND × RETURN			−0.024	−0.24
S × MAND × VOLATILITY			0.576	1.27
S × MAND × PROF			−1.021	−1.42
Intercept	−3.141 ***	−7.49	−3.105 ***	−5.04
Industry Effects	YES		YES
Year Effects	YES		YES
Pseudo R2	0.459		0.465
Observations	10,785		10,785

Notes: Table 3 presents regression results estimating the likelihood of a modified audit opinion (MO), equal to 1 for unqualified opinions with harmful explanatory language, qualified opinions, adverse opinions, or disclaimers, and 0 otherwise. Column 1, estimated from Equation (1), includes control variables and interactions with auditor switching (S) using the pooled sample. Column 2, based on Equation (2), introduces interactions with a mandatory regime indicator (MAND) to explicitly separate the voluntary and mandatory samples. The voluntary sample estimates in column 2 are used to simulate the probability of a modified opinion for mandated firms under two counterfactual voluntary scenarios. Standard errors are clustered at the firm level. *, **, and *** denote two-tailed statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Appendix A defines all variables.

presents results from logistic regressions where the dependent variable equals one for modified audit opinions. Column 1 estimates a pooled model. As expected, prior modification (MOLAG) is highly predictive (β = 2.034, z = 49.35), and firms with high leverage (LEV) and book-to-market ratio (BM) or low profitability (PROF) and operating cash flows (OCF) are more likely to receive modified opinions. The coefficient on S × MOLAG is negative (β = −0.719, z = −7.81), indicating that switching auditors reduces the persistence of prior modifications.

Column 2 adds interaction terms with MAND to differentiate voluntary from regulated engagements. Results remain consistent: S×MOLAG remains negative and significant (β = −0.470, z = −3.59), while MAND×MOLAG is positive and significant (β = 0.179, z = 2.30), suggesting increased opinion stickiness under mandatory retention. The three-way interaction S × MAND × MOLAG is negative and significant (β = −0.506, z = −2.56), indicating that mandated auditor changes may attenuate this persistence.

4.5. Identifying Disguised Opinion Shopping

As shown in

Table 4. Observed mandatory rotation or retention vs. predicted preferences for audit firm switching or retention.

Panel A: Observed Mandatory Rotation or Retention to Predicted Preference
	Predicted
Observed	Switching Preferred (S = 1)	Retention Preferred (S = 0)	Total
Mandatory Rotation (MAND = 1 & S = 1)	180	314	494
Mandatory Retention (MAND = 1 & S = 0)	2932	4030	6962
Total	3112	4344	7456
Panel B: Identification of Disguised Opinion Shopping
Aligned =	180 + 4030 =	4210	56.5%
Misaligned =	314 + 2932 =	3246	43.5%
Total =		7456	100%

Notes: Table 4 presents the distribution of mandated firms by whether their observed auditor arrangement corresponds to the auditor choice predicted to reduce the likelihood of a modified audit opinion. These predictions are based on voluntary sample estimates from Table 3 and compare the outcomes under two counterfactual scenarios: voluntary retention and voluntary switching. If the observed mandate aligns with the option yielding the lower modified audit opinion, the firm is classified as a disguised opinion shopper (DOS = 1); otherwise, it is classified as a non-opinion shopper (DOS = 0). The top panel provides a detailed cross-tabulation, while the bottom panel summarizes matched and mismatched classifications.

, firm-year observations are classified by comparing the mandated auditor outcome with the predicted preference. If the observed arrangement—retention or switch—matches the counterfactual associated with the lower probability of modification, the firm is coded as a disguised opinion shopper (DOS = 1). Otherwise, it is treated as a non-opinion shopper (DOS = 0). Among 7456 firm-years, 4210 (56.5%) exhibit alignment: 180 of 494 mandatory rotation cases (36.4%) and 4030 of 6962 mandatory retention cases (57.9%) match the firm’s inferred preference. The remaining 3246 observations (43.5%) reflect preference misalignment.

Figure 2 illustrates the empirical framework used to infer firms’ counterfactual preferences for switching or retention and identify disguised opinion shopping (DOS). Predicted probabilities of modified opinions under hypothetical discretion are estimated using voluntary engagements. DOS is classified when a firm’s mandated outcome (rotation or retention) aligns with its inferred preference. The framework also outlines hypothesis tests of audit outcomes by DOS status and mandate type (H1–H3), as presented in the following section.

5. Empirical Analyses

5.1. Descriptive Statistics

Table 5. Univariate differences in DOS (1 vs. 0) within mandatory rotation and retention samples.

	(1) Mandatory Rotation Sample					(2) Mandatory Retention Sample
	Mean in DOS = 0	Mean in DOS = 1	Diff. in Means	t-Stat	p-Value	Mean in DOS = 0	Mean in DOS = 1	Diff. in Means	t-Stat	p-Value
Dependent Variables
MO	0.32	0.59	−0.27	−6.1 ***	(0.00)	0.84	0.34	0.49	48.87 ***	(0.00)
DA	0.03	−0.07	0.10	4.58 ***	(0.00)	−0.05	0.02	−0.07	−16.2 ***	(0.00)
FSD_SCORE	3.55	3.29	0.26	2.42 **	(0.02)	3.42	3.54	−0.12	−4.21 ***	(0.00)
VOL_SWITCH	0.43	0.37	0.06	1.28	(0.20)	0.12	0.09	0.03	4.08 ***	(0.00)
Independent Variables
MOLAG	0.21	0.91	−0.70	−22 ***	(0.00)	0.95	0.27	0.68	83.5 ***	(0.00)
SIZE	18.63	18.56	0.08	0.68	(0.50)	18.84	18.27	0.57	15.54 ***	(0.00)
LIQ	2.50	2.09	0.41	1.53	(0.13)	2.34	2.58	−0.24	−3.25 ***	(0.00)
DE	0.10	0.12	−0.02	−2.1 **	(0.03)	0.15	0.10	0.05	14.60 ***	(0.00)
REC_INV	0.28	0.27	0.02	1.04	(0.30)	0.27	0.30	−0.03	−6.90 ***	(0.00)
ISSUANCE	0.41	0.41	0.01	0.18	(0.85)	0.52	0.46	0.06	5.29 ***	(0.00)
ZSCORE	3.02	1.49	1.53	3.99 ***	(0.00)	2.39	3.51	−1.12	−11.2 ***	(0.00)
TENURELAG	3.16	3.13	0.02	0.72	(0.47)	2.09	2.05	0.04	1.99 **	(0.05)
BIGN	0.46	0.47	−0.00	−0.04	(0.97)	0.60	0.54	0.05	4.57 ***	(0.00)
SPECIALIST	0.04	0.10	−0.06	−2.3 **	(0.02)	0.19	0.11	0.08	8.69 ***	(0.00)
CI_FEE	0.06	0.06	−0.00	−0.05	(0.96)	0.04	0.05	−0.01	−4.28 ***	(0.00)
DELAY	3.91	3.96	−0.05	−1.59	(0.11)	3.80	3.88	−0.08	−9.86 ***	(0.00)
LISTINGAGE	13.97	13.78	0.19	0.25	(0.80)	13.34	11.02	2.31	10.19 ***	(0.00)
DELIST	0.12	0.26	−0.14	−3.6 ***	(0.00)	0.20	0.18	0.01	1.49	(0.14)
RETURN	−0.16	−0.34	0.18	2.77 ***	(0.01)	0.12	0.30	−0.17	−7.25 ***	(0.00)
VOLATILITY	0.68	0.71	−0.03	−1.56	(0.12)	0.63	0.66	−0.04	−3.29 ***	(0.00)
INDBOARD	0.23	0.22	0.01	0.58	(0.56)	0.21	0.19	0.02	5.42 ***	(0.00)
FOREIGN	0.23	0.33	−0.10	−3.2 ***	(0.00)	0.29	0.25	0.04	5.58 ***	(0.00)
SQSEG	1.49	1.53	−0.04	−0.96	(0.34)	1.49	1.46	0.03	3.07 ***	(0.00)
LAH	6.55	6.53	0.01	0.25	(0.80)	6.43	6.27	0.15	5.97 ***	(0.00)
EST_AGE	28.21	28.31	−0.10	−0.08	(0.93)	27.66	24.45	3.21	9.09 ***	(0.00)
OCF	0.02	0.06	−0.05	−3.6 ***	(0.00)	0.06	0.01	0.05	14.08 ***	(0.00)
IND_GROWTH	0.15	0.17	−0.02	−1.72 *	(0.09)	0.11	0.11	0.00	1.01	(0.31)
LEV	0.41	0.52	−0.11	−4.1 ***	(0.00)	0.50	0.40	0.10	17.12 ***	(0.00)
TENURE	1.00	1.00				5.27	5.03	0.24	2.56 **	(0.01)
CASH	0.07	0.06	0.01	2.16 **	(0.03)	0.06	0.07	−0.01	−4.09 ***	(0.00)
PROF	−0.01	−0.22	0.21	4.73 ***	(0.00)	−0.10	0.00	−0.10	−12.5 ***	(0.00)
LOSS	0.32	0.42	−0.10	−2.2 **	(0.03)	0.30	0.29	0.01	0.90	(0.37)
GROWTH	0.18	0.11	0.07	1.95 *	(0.05)	0.05	0.15	−0.10	−11.7 ***	(0.00)
ACQUIRE	0.08	0.11	−0.03	−1.04	(0.30)	0.11	0.08	0.04	4.89 ***	(0.00)
CFEARLY	0.53	0.45	0.07	1.41	(0.16)	0.36	0.45	−0.09	−6.77 ***	(0.00)
CFMATURE	0.26	0.37	−0.11	−2.2 **	(0.03)	0.43	0.30	0.13	9.56 ***	(0.00)
MODOPIN	0.87	0.92	−0.04	−1.55	(0.12)	0.91	0.57	0.34	36.70 ***	(0.00)
SHORT_TEN	1.00	1.00				0.46	0.48	−0.02	−1.56	(0.12)
N	314	180				2932	4030

Notes: Table 5 reports univariate differences in key variables between disguised opinion shoppers (DOS = 1) and non-opinion shoppers (DOS = 0), separately for the mandatory rotation sample (Column 1) and the mandatory retention sample (Column 2). *, **, and *** denote two-tailed statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Appendix A defines all variables.

summarizes descriptive statistics comparing firm-year observations classified as disguised opinion shoppers (DOS = 1) versus non-opinion shoppers (DOS = 0), separately for the mandatory rotation and mandatory retention subsamples. These comparisons offer initial evidence on how alignment between mandated auditor outcomes and firms’ counterfactual preferences relates to audit results, financial profiles, and governance structures.

In the mandatory rotation sample, firms identified as disguised opinion shoppers—those required and predicted to switch auditors—tend to exhibit weaker fundamentals. They are more likely to report operating losses (LOSS), lower profitability (PROF), higher leverage (LEV), and lower Altman Z-scores (ZSCORE). Stock performance is also poorer, with more negative returns (RETURN) and an increased likelihood of delisting (DELIST). Audit outcomes reflect this adversity: disguised shoppers are more likely to receive modified audit opinions (MOs), but report lower discretionary accruals (DAs) and financial statement divergence scores (FSD_SCORE), suggesting more conservative reporting. These firms also operate in less robust audit environments, with shorter lagged auditor tenure (TENURELAG) and lower use of industry specialists (SPECIALIST).

In the mandatory retention sample, the pattern reverses. Disguised opinion shoppers—firms required and predicted to retain their auditor—exhibit stronger financial and market characteristics. They report higher profitability (PROF), lower distress (higher Z-scores), and better stock returns (RETURN). While they hold less long-term debt (DE), they maintain higher total leverage (LEV). Audit outcomes are more favorable: these firms are less likely to receive modified opinions (MOs) and report higher discretionary accruals (DAs) and divergence scores (FSD_SCORE), suggesting increased flexibility in financial reporting. Moreover, they are more likely to voluntarily retain their auditor under non-mandated conditions (VOL_SWITCH).

Governance indicators also differ. Aligned firms under retention mandates are less likely to hire Big N auditors (BIGN) or industry specialists (SPECIALIST), exhibit lower board independence (IND_BOARD), and have shorter auditor tenure (TENURE)—suggesting weaker external oversight. These firms also tend to be smaller (SIZE), younger (EST_AGE and LISTINGAGE), and less active in capital markets (ISSUANCE). In contrast, non-opinion shoppers (DOS = 0) are generally more mature and complex: they exhibit higher operating cash flows (OCF), greater foreign revenue exposure (FOREIGN), and more organizational segments (SQSEG).

Overall, these descriptive patterns indicate that alignment under mandatory retention is associated with more favorable audit outcomes and looser oversight environments, consistent with reduced auditor independence. In contrast, alignment under mandatory rotation is not clearly beneficial, possibly due to knowledge disruption and auditor learning costs that dilute any strategic advantages.

5.2. Disguised Opinion Shopping and Audit Opinion Outcomes

To assess whether disguised alignment between clients and auditors results in more favorable audit opinions, the analysis begins by presenting descriptive evidence. Figure 3 presents the mean modified audit opinions for firms classified as DOS = 1 and DOS = 0 across the years surrounding auditor rotation, with the rotation event occurring at Year 1. The analysis is restricted to firms that undergo mandatory auditor rotation, allowing for a direct comparison of audit opinions issued by predecessor and successor auditors over their respective tenures within the sample period. Observations preceding Time 1 correspond to audit opinions issued by the predecessor auditor, while those following Time 1 reflect opinions issued by the successor auditor.

Following the auditor rotation at Year 1, distinct patterns emerge between the two groups. Firms with DOS = 1 exhibit a sharp decline in the incidence of modified audit opinions immediately after rotation, as indicated by the downward red arrow. In contrast, firms with DOS = 0 display an increase in modified audit opinions post-rotation, as reflected by the upward blue arrow. This divergence suggests that successor auditors may respond differently depending on whether the engagement reflects underlying alignment with client preferences—raising the possibility that DOS firms secure more favorable outcomes by influencing the auditor’s opinion formation process.

To examine this possibility more formally, logistic regressions are estimated where the dependent variable MO equals one if a firm receives a modified opinion in a given year. The primary independent variable is an indicator of engaging in DOS. The analysis also estimates a specification that disaggregates DOS into two distinct types: DOS_SWITCH, capturing alignment under mandatory rotation, and DOS_RETAIN, capturing alignment under mandatory retention.

$${\mathrm{MO}}_{\mathrm{i},\mathrm{t}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{DOS}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{X}}_{\mathrm{i},\mathrm{t}}+\mathrm{Year}+\mathrm{Industry}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(3)

Following prior literature (e.g., Lee et al., 2024), all regressions include controls for prior audit opinion, financial condition, governance structure, and audit engagement characteristics—such as firm size, leverage, liquidity, distress risk, operational complexity, auditor type, and engagement tenure. Year and industry fixed effects are included, and standard errors are clustered at the firm level.

Table 6. Disguised opinion shopping and audit opinion outcomes.

	Full Mandate Sample				Mandatory Rotation		Mandatory Retention
	(1)		(2)		(3)		(4)
Dep. =	MO		MO		MO		MO
Variable	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat
DOS	−0.254 ***	−4.82			−0.070	−0.29	−0.214 ***	−3.40
DOS_SWITCH			−0.675 ***	−4.32
DOS_RETAIN			−0.157 **	−2.47
MOLAG	1.889 ***	36.30	1.971 ***	31.51	1.272 ***	5.47	1.982 ***	32.98
S	−0.335 ***	−3.32	−0.095	−0.73
SIZE	0.063 **	2.43	0.069 ***	2.64	0.232 **	2.51	0.037	1.59
LIQ	−0.001	−0.11	0.001	0.14	−0.045	−1.36	0.007	0.66
DE	0.314	1.59	0.322	1.64	0.807	0.85	0.476 **	2.55
REC_INV	−0.131	−0.92	−0.151	−1.05	−0.971 *	−1.75	0.122	0.84
ISSUANCE	−0.038	−0.92	−0.038	−0.91	−0.176	−1.07	−0.048	−1.16
ZSCORE	−0.038 ***	−4.44	−0.039 ***	−4.55	−0.031	−0.97	−0.044 ***	−5.07
TENURELAG	0.057 **	2.12	0.058 **	2.17	−0.157	−0.74	0.135 ***	5.29
BIGN	−0.772 ***	−15.30	−0.754 ***	−14.95	−1.656 ***	−7.88	−0.609 ***	−12.20
SPECIALIST	0.023	0.38	0.027	0.44	0.345	1.08	−0.006	−0.10
CI_FEE	−0.057	−0.36	−0.063	−0.39	−0.434	−0.52	−0.031	−0.19
DELAY	0.095	1.44	0.098	1.49	0.396	1.36	−0.110 *	−1.71
LISTINGAGE	−0.003	−1.11	−0.003	−1.16	−0.017	−1.48	−0.004 *	−1.74
DELIST	0.161 ***	2.66	0.169 ***	2.76	0.387	1.64	0.166 ***	2.86
RETURN	−0.009	−0.38	−0.014	−0.61	−0.071	−0.69	−0.056 **	−2.37
VOLATILITY	0.600 **	2.51	0.591 **	2.48	1.283 ***	2.79	0.673 ***	2.88
INDBOARD	−0.053	−0.35	−0.068	−0.44	0.399	0.65	−0.400 ***	−2.66
FOREIGN	0.071	0.92	0.078	1.01	0.391	1.32	0.059	0.77
SQSEG	0.153 ***	3.42	0.152 ***	3.39	0.181	1.07	0.131 ***	2.92
Intercept	−2.605 ***	−3.32	−2.842 ***	−3.60	−6.738 ***	−2.98	−1.072	−1.46
Industry Effects	YES		YES		YES		YES
Year Effects	YES		YES		YES		YES
Pseudo R2	0.491		0.492		0.408		0.473
Observations	7456		7456		494		6962

Notes: Table 6 investigates the association between disguised opinion shopping (DOS = 1) and the likelihood of receiving a modified audit opinion. Columns 1 and 2 report results for the full mandate sample, while Columns 3 and 4 focus on the mandatory rotation and mandatory retention subsamples, respectively. Column 2 replaces the overall DOS indicator with separate indicators for disguised switching (DOS_SWITCH) and disguised retention (DOS_RETAIN). All regressions control for firm, engagement, and audit characteristics, with industry and year fixed effects. Standard errors are clustered at the firm level. *, **, and *** denote two-tailed significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Appendix A defines all variables.

presents the regression results. In Column 1, the coefficient on DOS is negative and statistically significant (−0.254, z = −4.82), indicating that firms with aligned auditor preferences are significantly less likely to receive a modified audit opinion. This result provides strong support for H1. Column 2 separates DOS into its two underlying components, DOS_SWITCH and DOS_RETAIN, both of which are negatively associated with the likelihood of modification. However, subsample analyses in Columns 3 and 4 reveal that the association is statistically significant only under mandatory retention (−0.214, z = −3.40), where DOS is negatively related to modification outcomes, consistent with H3. In contrast, under mandatory rotation, DOS status has no discernible effect on audit opinions, in line with H2.

The coefficients on control variables are broadly consistent with prior literature. The strongest predictor of a current-year modified opinion (MO) is MOLAG, an indicator for a prior-year modification, consistent with the well-documented persistence of audit opinions. ZSCORE, a measure of financial health, is negatively associated with the likelihood of modification, suggesting that more financially stable firms face lower audit risk. The coefficient on SIZE is positive and significant, indicating that larger firms are more likely to receive modified opinions, possibly due to increased operational complexity or greater exposure to regulatory scrutiny. DELIST is positively associated with MO, suggesting that firms facing delisting risk are more likely to receive adverse audit outcomes. Firms audited by Big N auditors are significantly less likely to receive modified opinions, consistent with prior evidence on audit quality differentials. Finally, firms with greater market volatility (VOLATILITY) and more complex operations (SQSEG) are more likely to be qualified, reflecting the increased risk and information asymmetry associated with these attributes.

Taken together, the results in Table 6 indicate that the likelihood of receiving a modified opinion is significantly lower for firms engaging in disguised opinion shopping (DOS = 1). If managers in such firms are motivated to obtain more favorable audit outcomes and succeed in engaging auditors who are more accommodating to their reporting preferences, these firms should exhibit a lower likelihood of receiving modified opinions relative to non-opinion-shopping firms. Consistent with this prediction, the evidence points to opportunistic auditor selection conducted under the appearance of regulatory compliance. Importantly, this effect is concentrated under mandatory retention, where auditor continuity is preserved, suggesting that the consequences of disguised alignment are more pronounced when ongoing auditor–client relationships enable accommodative dynamics to develop or persist.

5.3. Disguised Opinion Shopping and Audit Quality

Disguised opinion shopping may not only lead to more favorable audit opinions but also erode audit quality more broadly. Auditors who issue unqualified opinions when a going-concern modification is warranted may compromise their independence, enabling opportunistic financial reporting. Consistent with this concern, DeFond et al. (2025) document that firms selecting successor auditors more tolerant of income-increasing changes to accounting estimates report more favorable estimates and exhibit broader declines in audit quality.

Even in the absence of opinion shopping, prior research highlights quality risks associated with auditor turnover. Newly appointed auditors—especially those subject to regulatory rotation—often lack client-specific knowledge during the initial engagement period, heightening their reliance on management-provided estimates and impairing audit effectiveness. DeFond et al. (2025) further argue that, due to limited experience with the client, newly appointed auditors are particularly susceptible to being fooled by management during the early stages of the engagement. Arruñada and Paz-Ares (1997) also describe a pattern of “risky client shifting,” whereby outgoing auditors disengage from problematic clients without addressing lingering issues. Auditors near the end of their tenure may become less vigilant, increasing the likelihood that such issues go unaddressed. Consequently, successor auditors may face a steeper learning curve in identifying and addressing inherited risks.

To examine whether DOS is associated with lower audit quality, this study analyzes two widely used proxies: signed performance-matched discretionary accruals (DAs; Kothari et al., 2005) and financial statement divergence scores (FSD_SCORE; Amiram et al., 2015). Higher values of both measures indicate lower audit quality. DA captures accrual-based earnings management, while FSD_SCORE reflects statistical deviations from Benford’s Law and is interpreted as evidence of irregular or manipulated financial reporting (e.g., Minichilli et al., 2022; Gupta et al., 2020). I estimate the following model based on the specification in Equation (3) of Kwon et al. (2014):

$$\begin{array}{l}{\mathrm{DA}}_{\mathrm{i},\mathrm{t}}/\mathrm{FSD}\_{\mathrm{SCORE}}_{\mathrm{i},\mathrm{t}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{DOS}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{LTA}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{BIGN}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_4{\mathrm{EST}\_\mathrm{AGE}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_5{\mathrm{OCF}}_{\mathrm{i},\mathrm{t}}\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+{\mathsf{\beta }}_6{\mathrm{IND}\_\mathrm{GROWTH}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_7{\mathrm{LIQ}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_8{\mathrm{DE}}_{\mathrm{i},\mathrm{t}}+\mathrm{Year}+\mathrm{Industry}+{\mathsf{\epsilon }}_{\mathrm{it}}\end{array}$$

(4)

Table 7. Disguised opinion shopping and audit quality.

	Full Mandate Sample				Mandatory Rotation		Mandatory Retention		Full Mandate Sample				Mandatory Rotation		Mandatory Retention
	(1)		(2)		(3)		(4)		(5)		(6)		(7)		(8)
Dep.=	DA		DA		DA		DA		FSD_SCORE		FSD_SCORE		FSD_SCORE		FSD_SCORE
Variable	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat
DOS	0.044 ***	10.26			−0.029	−1.09	0.056 ***	13.12	0.101 ***	3.31			−0.044	−0.36	0.119 ***	3.67
DOS_SWITCH			−0.008	−0.61							−0.110	−1.15
DOS_RETAIN			0.047 ***	10.84							0.113 ***	3.65
LAH	−0.007 **	−2.40	−0.006 **	−2.32	0.010	0.35	−0.007 **	−2.53	0.027	1.43	0.028	1.48	−0.231 *	−1.79	0.032 *	1.68
SIZE	0.019 ***	9.42	0.019 ***	9.41	−0.000	−0.03	0.020 ***	10.04	−0.047 ***	−3.18	−0.047 ***	−3.14	0.021	0.30	−0.045 ***	−2.99
BIGN	0.000	0.02	−0.000	−0.04	0.022	0.78	0.001	0.17	−0.146 ***	−4.59	−0.147 ***	−4.62	−0.105	−0.85	−0.146 ***	−4.4
EST_AGE	0.000 **	2.17	0.000 **	2.37	0.000	0.43	0.000 **	2.31	0.001	0.71	0.001	0.83	−0.006	−1.27	0.001	0.98
OCF	−0.676 ***	−46.49	−0.670 ***	−45.88	−0.890 ***	−9.33	−0.652 ***	−45.99	0.367 ***	2.87	0.393 ***	3.06	−0.089	−0.18	0.415 ***	3.1
IND_GROWTH	0.011	0.59	0.012	0.64	0.006	0.04	0.020	1.03	−0.667 ***	−4.92	−0.643 ***	−4.74	−1.388 **	−2.34	−0.647 ***	−4.5
LIQ	−0.001	−1.47	−0.001	−1.28	−0.003	−0.72	−0.001	−0.79	−0.001	−0.25	−0.001	−0.14	0.040 *	1.67	−0.004	−0.61
LEV	−0.064 ***	−5.51	−0.059 ***	−5.07	−0.033	−0.53	−0.057 ***	−4.93	−0.208 **	−2.27	−0.190 **	−2.06	−0.125	−0.40	−0.195 **	−2.01
TENURE	−0.000	−0.05	−0.000	−0.55			0.000	0.26	−0.011 ***	−2.98	−0.013 ***	−3.35			−0.012 ***	−3.16
Intercept	−0.314 ***	−9.01	−0.319 ***	−9.16	0.017	0.07	−0.336 ***	−10.09	4.409 ***	18.13	4.380 ***	18.00	4.991 ***	4.82	4.311 ***	17.55
Industry Effects	YES		YES		YES		YES		YES		YES		YES		YES
Year Effects	YES		YES		YES		YES		YES		YES		YES		YES
Adj. R-squared	0.288		0.290		0.318		0.312		0.039		0.040		0.200		0.039
Observations	7456		7456		494		6962		7456		7456		494		6962

Notes: Table 7 examines the association between disguised opinion shopping and audit quality, measured by discretionary accruals (DAs) and financial statement divergence scores (FSD_SCORE). Columns 1–4 use DAs, calculated using the performance-matched accrual model of Kothari et al. (2005). Columns 5–8 use FSD_SCORE, following the measure developed by Amiram et al. (2015). Results are presented for the full mandate sample, as well as for the mandatory rotation and mandatory retention subsamples. Standard errors are clustered at the firm level. *, **, and *** indicate two-tailed significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Appendix A defines all variables.

presents the regression estimates examining the association between disguised opinion shopping and audit quality. In the full sample, DOS is positively and significantly associated with performance-matched discretionary accruals (coefficient = 0.044, z = 10.26) in Column 1, consistent with higher levels of earnings management among opinion-shopping firms. Column 2 decomposes the DOS indicator into two components: DOS_RETAIN and DOS_SWITCH. The coefficient on DOS_RETAIN is positive and significant (coefficient = 0.047, z = 10.84), whereas the coefficient on DOS_SWITCH is statistically insignificant. These results suggest that the observed decline in audit quality is driven by firms that retain their incumbent auditors under mandatory retention, rather than those that switch auditors under mandatory rotation.

The results using FSD_SCORE as the dependent variable yield a similar pattern. In the full sample (Column 5), DOS is positively and significantly associated with higher FSD scores (coefficient = 0.101, z = 3.31), indicating a greater likelihood of statistical irregularity in financial reporting. Column 6 shows that this effect is again concentrated among DOS_RETAIN firms (coefficient = 0.113, z = 3.65), with no significant association observed for DOS_SWITCH.

Subsample analyses further corroborate these findings. Among firms subject to mandatory retention, DOS is significantly associated with both higher discretionary accruals (coefficient = 0.056, z = 13.12) in Column 4 and higher FSD scores (coefficient = 0.119, z = 3.67) in Column 8. In contrast, among firms subject to mandatory rotation, DOS is not significantly related to either outcome measure in Columns 3 and 7. These patterns suggest that the adverse audit quality implications of disguised alignment are concentrated in settings where auditor continuity is preserved.

Control variables behave largely as expected and are broadly consistent with prior literature (e.g., Kwon et al., 2014). LAH is negatively associated with DA, in line with the view that greater audit effort mitigates earnings management, but is not significant in the FSD regressions. SIZE is positively associated with DA and negatively associated with FSD_SCORE, implying that larger firms exhibit greater accrual-based earnings management but fewer statistical anomalies. BIGN is insignificant in the DA models but significantly negative in the FSD models, consistent with prior evidence that Big N auditors are associated with higher-quality reporting. OCF is negatively associated with both DA and FSD_SCORE, suggesting that firms with stronger operating cash flows are less likely to engage in aggressive financial reporting. LEV is also negatively associated with both outcomes, consistent with the notion that more highly leveraged firms are subject to greater monitoring and exhibit more conservative reporting behavior.

Overall, the results in Table 7 provide consistent evidence that disguised opinion shopping—particularly when it arises in auditor retention—is associated with diminished audit quality. These results align with the predictions of multi-period audit frameworks (e.g., Dye, 1993; Watts & Zimmerman, 1983); auditors tend to exert less effort and show more leniency when they face little risk of dismissal and weak oversight.

The absence of similar effects under mandatory rotation suggests that auditor turnover disrupts client-specific knowledge and impacts the effectiveness of disguised opinion shopping. This finding is consistent with prior evidence that mandatory rotation may fail to enhance—and may even impair—audit quality due to auditor unfamiliarity and engagement discontinuity (Kalanjati et al., 2019; Kwon et al., 2014). Such declines are particularly pronounced in the early years of mandatory tenure, when steep learning curves limit auditor effectiveness—a pattern documented across institutional settings, including Italy (Cameran et al., 2015, 2016), the Netherlands (de Jong et al., 2020), and India (Dayanandan & Kuntluru, 2023).

5.4. Voluntary Auditor Selection in the Post-Mandate Period

To examine whether firms flagged for disguised opinion shopping (DOS) are more likely to retain their auditors when firms regain discretion over auditor selection, this study analyzes voluntary auditor switching decisions (VOL_SWITCH) in the three-year period following a firm’s initial mandatory assignment—whether through rotation or retention. This setting provides a useful test of whether favorable audit outcomes secured under regulatory alignment are rewarded with continued engagements once the auditor selection constraint is lifted. Following prior research on auditor retention (e.g., Brown & Knechel, 2016; Hunt et al., 2021), I estimate a logistic regression model in which the dependent variable equals one if the firm voluntarily switches auditors in the post-mandate period and zero otherwise. To isolate discretionary choices, I exclude firms that are newly subject to rotation at this juncture.

$${\mathrm{VOL}\_\mathrm{SWITCH}}_{\mathrm{i},\mathrm{t}}={\mathsf{\beta }}_0 + {\mathsf{\beta }}_1{\mathrm{DOS}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{X}}_{\mathrm{i},\mathrm{t}}+\mathrm{Year}+\mathrm{Industry}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(5)

Table 8. Voluntary auditor selection in the post-mandate period.

	Full Mandate Sample				Mandatory Rotation		Mandatory Retention
	(1)		(2)		(3)		(4)
Dep. =	VOL_SWITCH		VOL_SWITCH		VOL_SWITCH		VOL_SWITCH
Variable	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat
DOS	−0.343 ***	−3.31			−0.147	−0.94	−0.611 ***	−3.42
DOS_SWITCH			−0.233	−1.59
DOS_RETAIN			−0.440 ***	−3.23
SIZE	−0.078	−1.56	−0.088 *	−1.73	−0.117	−1.38	−0.051	−0.70
REC_INV	−0.336	−0.87	−0.339	−0.88	−0.440	−0.79	−0.154	−0.25
DA	0.826 *	1.75	0.927 *	1.91	−0.965	−0.99	0.827	0.91
CASH	−1.168 *	−1.70	−1.137 *	−1.65	−1.363	−1.08	−1.127	−1.17
PROF	−1.614 ***	−3.50	−1.546 ***	−3.37	−0.967 *	−1.78	−3.249 ***	−3.08
LOSS	0.008	0.05	0.025	0.17	−0.153	−0.76	−0.037	−0.14
GROWTH	−0.003	−0.02	−0.001	−0.01	0.017	0.07	−0.189	−0.79
ACQUIRE	−0.296	−1.59	−0.287	−1.53	−0.159	−0.46	−0.483 **	−1.97
CFEARLY	−0.192	−1.40	−0.181	−1.31	−0.302	−1.60	−0.013	−0.05
CFMATURE	0.123	0.81	0.124	0.82	0.066	0.31	0.168	0.66
MODOPIN	0.200	1.57	0.165	1.25	−0.098	−0.47	0.485 **	2.45
SPECIALIST	−0.097	−0.59	−0.100	−0.61	−0.479	−1.40	−0.095	−0.43
SHORT_TEN	−0.404 ***	−3.19	−0.455 ***	−3.38			−0.276 *	−1.67
BIGN	−0.159	−1.36	−0.159	−1.36	−0.278 *	−1.67	0.155	0.76
Intercept	1.410	1.38	1.659	1.59	1.534	0.94	1.045	0.68
Industry Effects	YES		YES		YES		YES
Year Effects	YES		YES		YES		YES
Pseudo R2	0.120		0.121		0.116		0.222
Observations	1034		1034		445		589

Notes: Table 8 examines the relationship between disguised opinion shopping and voluntary auditor switching after the mandate period, once auditor choice discretion is restored. The dependent variable, VOL_SWITCH, equals 1 if a firm voluntarily changes auditors post-mandate. Columns 1–2 use the full mandate sample, while Columns 3–4 report results separately for the mandatory rotation and retention subsamples. Column 2 replaces the overall DOS indicator with separate indicators for DOS_SWITCH and DOS_RETAIN. Standard errors are clustered at the firm level. *, **, and *** denote two-tailed significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Appendix A defines all variables.

reports the results. In Column 1, the coefficient on DOS is negative and statistically significant (−0.343, z = −3.31), indicating that firms identified as disguisers are less likely to initiate auditor changes once the regulatory constraint expires. This finding is consistent with the interpretation that favorable treatment received under mandated alignment is reciprocated through auditor retention when choice is restored.

In Column 2, I distinguish between firms that previously retained (DOS_RETAIN) versus switched (DOS_SWITCH) auditors during the mandate period. The negative association is concentrated in the DOS_RETAIN group (−0.440, z = −3.23), while the coefficient on DOS_SWITCH is negative but not statistically significant. These results suggest that firms that benefited from favorable audit outcomes under retention are particularly likely to preserve the incumbent auditor relationship.

Columns 3 and 4 present results for the mandatory rotation and mandatory retention subsamples, respectively. The negative association between DOS and voluntary switching is statistically significant only in the retention subsample (−0.611, z = −3.42), and not in the rotation subsample. These findings reinforce the argument that retention settings facilitate sustained alignment, as continuity allows the auditor–client relationship to mature and potentially entrench accommodative dynamics.

Control variables behave largely as expected (Brown & Knechel, 2016; Hunt et al., 2021). PROF is consistently negative and statistically significant across specifications, suggesting that more profitable firms are less inclined to alter established audit relationships. SHORT_TEN is also significantly negative, consistent with prior findings that firms are less likely to switch auditors early in the engagement, possibly due to initial relationship-specific investments or lack of switching incentives. SIZE is negatively associated with switching and marginally significant in Column 2, aligning with the view that larger firms may face higher switching frictions. CASH is also negatively associated with switching in both models, albeit at marginal significance levels, indicating that more liquid firms may perceive less urgency to revise audit arrangements.

Overall, the results in Table 8 indicate that firms engaged in disguised opinion shopping—particularly those that retained their auditors under mandatory retention—are significantly less likely to switch auditors once discretion is restored. These findings suggest that regulatory alignment via retention may enable firms to entrench auditor–client relationships that accommodate managerial preferences, potentially extending the influence of opinion shopping behavior beyond the mandate period.

These findings are consistent with recent evidence that firms reward accommodating auditors. Ege and Stuber (2022) show that audit offices exhibiting greater lenience—by allowing managerial discretion in accounting estimates—are more likely to retain existing clients, even those not directly engaging in earnings manipulation. Similarly, Cowle et al. (2023) find that audit offices issuing a higher number of adverse internal control opinions experience lower client growth, particularly when the disclosures involve visible clients or entity-level control weaknesses. These offices are less likely to be selected by new clients, and regain market share only after reducing the frequency of adverse disclosures.

Extending this insight to the context of mandatory tenure rules, the evidence in Table 8 suggests that auditors who align with managerial preferences during the mandate—by issuing fewer modified audit opinions or tolerating more aggressive financial reporting—are rewarded through continued engagements once auditor choice is restored.

6. Additional Analyses

6.1. Sensitivity of Audit Quality Outcomes to Modified Opinion Classification

Audit opinion severity varies, and firms may be more motivated to avoid particularly damaging forms. To assess whether the main results depend on how modified opinions are defined,

Table 9. Sensitivity of audit quality outcomes to modified opinion classification.

	Exclude Related Party Transactions								Include Potentially Negative Explanatory Language
	(1)		(2)		(3)		(4)		(5)		(6)		(7)		(8)
Dep. =	DA		DA		FSD_SCORE		FSD_SCORE		DA		DA		FSD_SCORE		FSD_SCORE
Variable	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat
DOS	0.032 ***	7.19			0.043	1.31			0.047 ***	11.37			0.115 ***	3.82
DOS_SWITCH			−0.001	−0.14			−0.037	−0.51			0.003	0.21			−0.053	−0.57
DOS_RETAIN			0.037 ***	7.91			0.058 *	1.66			0.050 ***	11.83			0.125 ***	4.10
Controls	YES		YES		YES		YES		YES		YES		YES		YES
Industry Effects	YES		YES		YES		YES		YES		YES		YES		YES
Year Effects	YES		YES		YES		YES		YES		YES		YES		YES
Adj. R-squared	0.283		0.284		0.037		0.038		0.291		0.292		0.040		0.040
Observations	7456		7456		7456		7456		7456		7456		7456		7456

Notes: Table 9 reports sensitivity tests for the audit quality results in response to changes in the definition of modified opinions. Columns 1–4 exclude related-party transactions from the harmful opinion category, while Columns 5–8 include potentially negative explanatory language. Audit quality is measured using discretionary accruals (DAs) and financial statement divergence scores (FSD_SCORE). Standard errors are clustered at the firm level. * and *** indicate two-tailed significance at the 10 percent and 1 percent levels, respectively. Appendix A defines all variables.

re-estimates the audit quality models using two alternative classifications. Columns 1–4 narrow the scope of modified opinions by excluding audit reports that mention related-party transactions, while Columns 5–8 adopt a broader classification that includes additional potentially negative explanatory language.

Across both approaches, DOS remains positively and significantly associated with discretionary accruals (0.032 and 0.047; z = 7.19 and 11.37), confirming a consistent link between disguised opinion shopping and higher earnings management. The association with financial statement divergence (FSD_SCORE) is also positive and significant in the broader specification (0.115; z = 3.82). Disaggregating DOS shows that the results are driven by DOS_RETAIN, while DOS_SWITCH remains insignificant. Untabulated subsample analysis yields consistent patterns.

These findings confirm that the negative association between disguised opinion shopping and audit quality is not sensitive to how modified opinions are defined. The effect persists across both restrictive and expansive classifications and is concentrated among firms that retain their incumbent auditors.

6.2. Alternative Thresholds for Inferring Firms’ Auditor Preference: Switch vs. Retain

To evaluate the robustness of the audit quality results to alternative methods for inferring firms’ auditor preferences, a more conservative threshold is implemented for identifying disguised opinion shopping (DOS). Following Chen et al. (2016), a firm is classified as preferring either switching or retention only if the absolute difference in its predicted probabilities of receiving a modified opinion under the two counterfactual scenarios exceeds 1%. Firms falling within this ±1% range—those effectively indifferent between switching and retaining—are treated as non-opinion-shoppers in Columns 1–4 of Table 9, and entirely excluded from the estimation sample in Columns 5–8.

The results reported in

Table 10. Alternative thresholds for inferring firms’ auditor preference: switch vs. retain.

	Apply 1% Cut-Off								Drop Observations Below 1% Cut-Off
	(1)		(2)		(3)		(4)		(5)		(6)		(7)		(8)
Dep. =	DA		DA		FSD_SCORE		FSD_SCORE		DA		DA		FSD_SCORE		FSD_SCORE
Variable	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat	Coeff	Z-Stat
DOS	0.042 ***	9.78			0.095 ***	3.16			0.043 ***	9.37			0.105 ***	3.19
DOS_SWITCH			−0.015	−1.09			−0.066	−0.68			−0.010	−0.72			−0.048	−0.49
DOS_RETAIN			0.045 ***	10.40			0.104 ***	3.41			0.046 ***	9.95			0.114 ***	3.42
Controls	YES		YES		YES		YES		YES		YES		YES		YES
Industry Effects	YES		YES		YES		YES		YES		YES		YES		YES
Year Effects	YES		YES		YES		YES		YES		YES		YES		YES
Adj. R-squared	0.287		0.289		0.038		0.038		0.315		0.317		0.040		0.040
Observations	7456		7456		7456		7456		6213		6213		6213		6213

Notes: Table 10 tests the robustness of audit quality results to the application of a 1 percent threshold when inferring auditor preference. Columns 1–4 reclassify firms whose auditor preference probabilities fall within a narrow ±1% band as non-opinion-shoppers while Columns 5–8 exclude these borderline observations from the sample. The dependent variables are discretionary accruals (DAs) and financial statement divergence scores (FSD_SCORE). Standard errors are clustered at the firm level. *** indicate two-tailed significance at the 1 percent level. Appendix A defines all variables.

demonstrate that the main findings are robust to this stricter preference filter. Across all specifications, DOS remains positively and significantly associated with lower audit quality. In the discretionary accruals (DAs) regressions in Columns 1 and 5, DOS firms exhibit significantly higher accruals, consistent with elevated earnings management. Similarly, in the financial statement divergence score (FSD_SCORE) regressions in Columns 3 and 7, DOS is associated with greater statistical irregularities, indicating higher reporting risk.

Disaggregating the DOS indicator offers further insight. In both the DA and FSD_SCORE regressions, only DOS_RETAIN—the subset of DOS firms that retained their incumbent auditors—is significantly associated with lower audit quality. DOS_SWITCH is not statistically significant in any specification. Untabulated subsample analyses support these conclusions.

Taken together, the findings in Table 10 confirm that the observed link between disguised opinion shopping and diminished audit quality is not sensitive to the classification threshold used to infer firms’ switching preferences. The effects persist under more conservative definitions of DOS and remain concentrated in mandatory retention.

7. Discussion and Contribution to Knowledge

This study provides the first empirical evidence on disguised opinion shopping. Prior research has focused on voluntary markets—where clients can freely dismiss and replace auditors (e.g., Lennox, 2000; Amin et al., 2021; DeFond et al., 2025)—but this study shifts attention to regulatory environments where auditor choice is constrained. In such settings, opinion shopping takes subtler forms, such as strategic alignment with accommodating auditors. This behavior is harder to observe, has not been previously examined, and—as the findings suggest—may pose even greater risks to audit quality.

Concerns about opinion shopping have resurfaced in regulatory deliberations regarding mandatory rotation (PCAOB, 2011). The PCAOB cautioned that increased competition for audit engagements—an intended consequence of mandatory rotation (Lennox, 2014)—could unintentionally reduce audit quality by incentivizing clients to seek more favorable audit opinions (PCAOB, 2011). Newton et al. (2016) provide empirical support for this concern, documenting a positive association between heightened audit market competition and the incidence of internal control opinion shopping. They further argue that mandatory rotation may exacerbate these dynamics by enhancing clients’ ability to exploit competitive pressure in pursuit of lenient audit outcomes. Moser (2021) offers a complementary theoretical perspective, demonstrating through a game-theoretic model that auditors operating in highly competitive markets may strategically lower audit effort or quality in order to appear more attractive to potential clients. These signaling incentives—particularly when fee pressure is present—may compromise auditor independence and facilitate opinion shopping, even within formally regulated environments.

Anecdotal evidence from Korea illustrates how these competitive dynamics may operate in practice. During the rollout period, large accounting firms actively invested in marketing efforts aimed at securing newly rotating clients. For example, Samjung-KPMG established a showroom to demonstrate its audit procedures and delivered customized pre-engagement presentations to prospective clients. A Samjung-KPMG partner observed that “the revenue rankings of accounting firms could shift dramatically depending on how many new clients each firm captures.” Similarly, Anjin-Deloitte launched a firmwide outreach campaign, stating, “We have launched an all-out marketing campaign to capitalize on this opportunity. We’ve already sent audit proposals to every company subject to mandatory rotation” (Korea Economic Daily, 2008). These efforts underscore how mandatory rotation may prompt auditors to signal flexibility and accommodation prior to engagement, laying the groundwork for disguised opinion shopping and reinforcing client expectations for favorable treatment—particularly among firms with heightened reporting risk.

Although no prior study investigates disguised opinion shopping in mandatory regimes, Chung et al. (2019) offer a useful conceptual reference. Their U.S.-based analysis focuses on distressed firms in a voluntary setting and defines opinion shopping narrowly as the avoidance of going-concern opinions. By contrast, this study draws on a broader sample of more than 7500 public firms in South Korea, where auditor tenure is governed by mandatory retention and rotation regulations. It expands the construct of opinion shopping to include standard modified opinions (qualified, adverse, and disclaimer) and unqualified opinions with harmful explanatory language dealing with more serious concerns, including going-concern uncertainties, financial distress, corporate reorganization, legal or regulatory violations, related-party transaction risks, and restatements of prior financial statements. While Chung et al. (2019) assess audit quality using misstatements and prediction errors, this study relies on alternative proxies—modified opinions, discretionary accruals, and financial irregularities.

Despite these design differences, both studies conclude that opinion shopping is associated with lower audit quality. Chung et al. (2019) report elevated misstatements primarily among firms that switch auditors. Consistent with H1, the present study finds that disguised opinion shopping is linked to audit deterioration. This effect is concentrated among firms that retain their auditors, who are significantly more likely to receive modified opinions, report aggressive accruals, and exhibit financial irregularities. These findings underscore how regulatory structure shapes the mechanisms through which opinion shopping affects audit outcomes.

Findings under mandatory rotation support H2, the auditor learning/competence hypothesis. Audit quality does not differ significantly between opinion-shopping and non-shopping switchers. This contrasts with voluntary settings, where audit quality deterioration is concentrated in auditor switching. As Chung et al. (2019, p. 102) explain, “successor auditors are incentivized to keep their new clients until they recover start-up costs and are thus more susceptible to client pressure,” and “if auditors are concerned about reputation damages borne by early termination of audit contract, the successor auditors subsequent to switching opinion shopping could be more vulnerable to the threats of dismissal”. Under mandatory rotation, these pressures are diminished: auditor turnover is routine, quasi-rents are truncated (Zhang, 1999), and reputational risks are reduced (Ruiz-Barbadillo et al., 2006).

More importantly, all newly appointed auditors face steep learning curves regardless of client incentives (Lennox, 2014). Audit quality becomes less responsive to opinion shopping motives because even compliant and non-opinion-shopping firms may experience degraded outcomes early in the engagement. Thus, null differences under rotation may not imply the absence of opportunism but reflect the structural limits on auditors’ ability to deliver differentiated audit quality in first-year audits.

By contrast, findings under mandatory retention support H3, the auditor closeness/independence hypothesis. Fixed-term engagements and limited auditor discretion reduce exit threats and weaken market discipline. In Korea, resignation due to client interference was not formally permitted until 2011, further insulating auditors. The finding that opinion-shopping firms are more likely to rehire the same auditor once discretion is restored suggests that clients reward accommodative behavior. These dynamics are consistent with multi-period audit models (e.g., Dye, 1993; Watts & Zimmerman, 1983), which predict lower effort and leniency when job security is assured and accountability is weak.

Taken together, these results show how regulatory design mediates the relationship between client incentives and audit outcomes. Audit quality is more favorable when mandate outcomes align with client preferences, supporting H1. Under mandatory rotation, auditor competence frictions reduce responsiveness to opinion-shopping incentives, supporting H2. Under mandatory retention, weak disciplinary forces and moral hazard increase the risk of auditor acquiescence, supporting H3 (Lennox, 2014).

Methodologically, this study also contributes a statistical detection model to identify firms engaging in disguised opinion shopping. In environments where switching no longer signals auditor independence concerns, such tools can aid oversight by regulators, audit committees, and institutional investors.

In sum, this study demonstrates that tenure-restricting regulations alter the form—but not the existence—of opinion shopping. By broadening the conceptual scope of opinion shopping, offering new evidence from a mandatory assignment regime, and introducing practical detection tools, this research deepens our understanding of how audit relationships evolve under constraint and where new risks to independence may emerge.

8. Conclusions

This study examines whether firms continue to engage in strategic auditor alignment—disguised opinion shopping—under mandatory retention and rotation rules that restrict auditor choice. While tenure mandates are designed to curb managerial discretion, regulators and scholars have long expressed concern that such rules may obscure, rather than eliminate, opportunistic behavior. To address this concern, the study develops a counterfactual framework using voluntary auditor engagements to estimate the likelihood of receiving a modified opinion from incumbent and successor auditors. These estimates are applied to mandated firms to infer whether switching or retaining would have reduced the probability of a modified opinion. A firm is classified as a disguised opinion shopper when its mandated outcome—rotation or retention—aligns with this inferred preference.

Disguised opinion shopping offers a way for firms to secure favorable audit outcomes while appearing compliant with regulation. Using Korean data from 2000 to 2009, this study finds that such firms are more likely to receive unmodified audit opinions and exhibit lower audit quality, as reflected in higher discretionary accruals and more frequent accounting irregularities. These effects are particularly evident under mandatory retention and disappear under mandatory rotation, where forced turnover appears to disrupt alignment. The evidence also suggests that firms benefitting from such alignment are more likely to reengage the same auditor once discretion is restored, consistent with a reward-for-accommodation mechanism.

These findings are consistent with the traditional view that opinion shopping reflects managerial opportunism. However, this study also suggests that not all auditor alignments under tenure mandates are necessarily opportunistic. While average effects point to lower audit quality, the regulatory structure may also incidentally align firms with auditors they would have preferred absent constraints, irrespective of intention. Therefore, detecting opportunism under formal compliance remains a nuanced empirical challenge.

A key contribution of this study is to show that pooling retention and rotation outcomes can obscure important heterogeneity in how tenure rules operate. This study shows that these mandates are not functionally equivalent: retention may entrench client–auditor relationships, while rotation imposes frictions that can limit opportunistic alignment. As tenure regulations continue to expand globally, future research could examine whether similar dynamics emerge under different institutional regimes or enforcement intensities.

Methodologically, this study advances a novel approach for identifying hidden preferences when traditional signals—such as voluntary auditor switches—are unavailable. This approach may inform future work on auditor–client dynamics under institutional constraints.

Several limitations apply to this study. First, it does not estimate the welfare effects of disguised opinion shopping or the broader economic cost of reduced audit quality. Second, as with any counterfactual approach, results may be sensitive to model specification and unobserved heterogeneity between mandated and voluntary firms. Third, the analysis is confined to a single country and regulatory period. As such, generalizing the findings beyond the Korean context should be approached with caution.

This study opens several avenues for future research into the implications of disguised opinion shopping. A particularly promising direction is to examine whether capital market participants—including investors, creditors, analysts, and credit rating agencies—recognize and respond to signals of strategic auditor alignment. For example, do firms flagged as disguised opinion shoppers experience valuation discounts, higher financing costs, altered analyst forecasts, or changes in credit ratings? Relatedly, future work could investigate how regulators and enforcement bodies interpret audit outcomes that reflect alignment with counterfactual auditor preferences.

Beyond market and regulatory responses, researchers may explore whether internal governance mechanisms—such as audit committee expertise, board independence, or ownership structure—moderate the likelihood or consequences of disguised alignment. Comparative studies across jurisdictions with varying legal regimes, enforcement intensity, audit market structures, and cultural norms could help identify institutional factors that shape the prevalence and consequences of strategic alignment.

Additional research could trace long-term firm outcomes—including financial restatements, litigation exposure, or audit fee patterns—to evaluate the downstream consequences of disguised opinion shopping, which may undermine auditor independence and diminish audit quality. Finally, qualitative and experimental studies, such as interviews or surveys with audit committee members, corporate executives, or auditors could offer deeper insights into how alignment decisions are made, justified, and governed under tenure mandates that formally constrain auditor choice.