Is the auditor’s industry specialization a reliable indicator of tax avoidance?
Andrew M. Bauer
Department of Accountancy
University of Illinois at Urbana-Champaign
Miguel Minutti-Meza
School of Business Administration
University of Miami
Andreya Marie Silva
School of Business Administration
University of Miami
August 2012
1
Is the auditor’s industry specialization a reliable indicator of tax avoidance?
ABSTRACT: This paper examines whether purchasing tax services from an auditor (APTS) with a large within-industry market share at city-level, a proxy for auditor expertise, affects the client’s level of tax avoidance. First, we examine cross-sectional differences in tax avoidance between clients of specialist and non-specialist auditors. Second, we demonstrate that, after matching clients of specialist and non-specialists along a number of client characteristics, there are no consistent differences in tax avoidance between clients of specialist and non-specialist auditors. Third, we perform a number of additional tests that confirm the results of our matched sample analyses, such as: using the FIN48-related reserves as a proxy for aggressive tax avoidance, including client fixed effects in our main models, and using alternative definitions of expertise at the office-level. Overall, we do not observe a consistent pattern of results indicating that purchasing APTS from an auditor with high market share results in comparatively greater tax avoidance. The evidence provided in this study suggests that an auditor’s within-industry market share is not a reliable indicator of tax avoidance. Nevertheless, we caution the reader that our results must be interpreted with due regard to their limitations and to the caveats of our methodology and highlight that our findings do not imply that industry knowledge is not important for auditors and tax advisors.
Keywords: tax avoidance, auditor industry specialization; effective tax rate; cash effective tax rate; matching.
Data availability: Data are publicly available from sources identified in the article.
1
Is the auditor’s industry specialization a reliable indicator of tax avoidance?
I. INTRODUCTION
As a response to concerns regarding independence, the Sarbanes-Oxley (SOX) Act restricts the types of non-audit services that a public company can purchase from its auditor.The auditor cannot provide most consulting services to the client, but SOX still permits the auditorto provide tax services (APTS) ifthe client’s audit committee grants approval. The passage of SOX motivated a sizeable number of clients to stop purchasing tax services from their external auditors (e.g., Omer et al.2006; Maydew and Shackelford 2007); nevertheless, over 58 percent of public companies purchased tax services from their external auditors after SOX and, on average, tax fees were close to20 percent of audit fees.[1]
The audit and tax literature has examined the consequences of purchasing tax services from the external auditor in addition to whether certain auditor’s characteristics, such as expertise, matter incrementally when purchasing joint audit and tax services. This paper focuses on whether purchasing tax services from an auditor with a large within-industry market share at the office level, a proxy for auditor expertise, affects the client’s level of tax avoidance.[2]Examining the benefits of receiving joint audit and tax services is important given the potential costs in terms of auditor independence. Moreover, asserting the effects of auditor industry specialization in the context of APTS is relevant for audit committees approving tax services from the auditor, to regulators concerned with issues of auditor independence, and to audit firms aiming to perform high-quality audits while maintaining their competitive position in each industry.
Theextant literature examining the effectsof APTS,suggests that there is a knowledge spillover between the audit and tax functions, which may increase the quality of the jointly provided services. Kinney et al. (2004) and Seetharaman et al. (2011),show thatboth accounting restatements and tax restatements are less likelyto occur for clients that receive APTS. Fortin and Pittman (2008) show that bondholders reward companies that pay proportionately more tax fees to their auditor with lower yield spreads.Gleason and Mills (2011) document that clients with APTS estimate full reserves for IRS disputes while clients without APTS require additional reserves in their estimate of tax expense. Conversely, there may be cases where APTS have negative consequences.For example, Cook et al. (2008) find that earnings management, through decreases in third-to-fourth quarter effective tax rate, is most prevalent in clients that pay comparatively higher tax fees to auditors.
In astudy closely related to this paper,McGuire et al. (2012) focus on differences between auditors that provide tax services. They suggest that within-industry market share may capture the size of the spillover effect between audit and tax services; and that this measure of industry specialization is a proxy for the tax expertise of audit firms. McGuire et al. (2012) find that the clients of industry specialist auditors have greater tax avoidance compared to the clients of non-specialist auditors. This study relies on arguments derived from the experimental and archival findings of the literature on auditor expertise and audit quality.
Experimental auditing research provides evidence that industry expertise enhances auditors’ judgements. The findings of prior studies suggest that knowledge of the industry may increase audit quality via improving the accuracy of error detection (Solomon et al. 1999; Owhoso et al. 2002), enhancing the quality of the auditor’s risk assessment (Taylor 2000; Low 2004), and influencing the choice of audit tests and the allocation of audit hours (Low 2004). Archival auditing research has also examined the effects of auditor industry expertise; however, archival researchers cannot directly observe expertise at the firm, office, or auditor level. Consequently, this area of the literature has used each audit firm’s within-industry market share or auditor industry specialization as an indirect proxy for auditor expertise.[3]
In archival settings, measuring the effects of auditor industry expertise on tax avoidance is problematic because the proxies for industry specialization and tax avoidance are associated with underlying client characteristics. For example, large clients can have lower effective tax rates and industry specialists often audit large clients. In addition, researchers examining the impact of APTS face a significant data limitation. One such limitation is that disclosed tax fees do not distinguish between amounts paid for tax planning and other services such as tax compliance and M&Astructuring. Moreover, clients are not required to disclose tax fees paidforservicesprovided by external non-auditor consultants or internal personnel. Thus, a potentially significant portion of tax planning services remains unknown, raising questions about the validity of cross-sectional comparisons among APTS and non-APTS clients.
Anecdotal evidence from individual tax experts argues that as long as the client and tax consultant have strong sharing and communication of information the tax planning strategies will be highly effective, regardless of whether those services are APTS or non-APTS (Hansen 2006). Arguably, the focus on communication, due diligence, and controls has been heightened in the post-SOX era, implying comparable tax planning strategies between APTS and non-APTS clients. Furthermore, the bundling of audit and tax fees provides an audit firm the opportunity to discount audit services while charging a premium on tax services, often perceived as a value-added effort.Donohoe and Knechel (2009) provide evidence that tax aggressive clients purchasing APTS pay a smaller audit premium than tax aggressive clientsnot purchasing APTS. The existence of a bundle discount may result in a potentialidentification problem when using market share, based on audit fees or joint fees, as a measure of accumulation of expertise.
Our study examines the association between tax avoidance and APTS provided by industry specialist auditors, aiming to control for the impact of client characteristics and the possibility of purchasingnon-disclosed tax services from advisors that are not the external auditor. Throughout our analyses there are two hierarchical categories of clients, those that purchase APTS and a subgroup of clients that purchase APTS from a specialist auditor. In order to determine the effect of auditor specialization, weseek to compare treated and control groups that have similar client characteristics, ideally approximating experimental conditions.The objective is to match treatment and control observations on all relevant observable dimensions except for the treatment and outcome variables. Thus, the clients of specialist auditors have to be compared to a control group of other clients. This control group may consist of other clients that purchased APTS from non-specialist auditors or may consist of all other clients in the sample.
In the context of APTS, matching is useful for two primary reasons. First, we do not observe any tax fees paid to non-auditor consultants by both APTS and non-APTS clients. If it is more likely that a comparable peer-client will engage in similar tax planning, the adequate comparison or counterfactual for a client of a specialist auditor is not all otherclients,but rather a peer-matched client with a similar level of APTS. Second, the linear control approach, including client size, performance, growth, and other control variables in multivariate regression analyses, has significantpotential problems such as selection biasand model misspecification. The literature on matching has documented that this technique is useful to mitigate both selection bias and model misspecification (e.g., Rosenbaum and Rubin 1983; Ho et al. 2007).
Our study unfolds in three dimensions. First, we examine cross-sectional differences in tax avoidance between clients of specialist and non-specialist auditors. We employ four proxies for tax avoidance: effective tax rate (ETR), cash effective tax rate (CETR), book-tax differences (BTD), and discretionary tax (DTAX); and use two definitions of specialization based on tax fees and joint audit and tax fess as proxies for expertise. Second, we demonstrate that, after matching clients of specialist and non-specialists along a number of client characteristics, there are no consistent differences in tax avoidance between clients of specialist and non-specialist auditors. Third, we perform a number of additional tests that confirm the results of our matched sample analyses, such as: using the FIN48-related reserves as a proxy for aggressive tax avoidance,including client fixed effects in our main models,and using alternative definitions of expertise based on the size and importance of the tax function at the office-level.
Throughout our analyses we do not observe a consistent pattern of results indicating that purchasing APTS from an auditor with high market share results in comparatively greater tax avoidance. The combined evidence provided in this study suggests that an auditor’s within-industry market share is not a reliable indicator of tax avoidance. Moreover, the extant empirical methodology may not fully parse out the confounding effects of client characteristics in tests of auditor industry specialization and tax avoidance.
We caution the reader that our results must be interpreted with due regard to their limitations and to the caveats of our methodology. Although they are appropriate for estimating causal effects, matching models have intrinsic restrictions, resulting from a trade-off between internal and external validity. In addition, our conclusions are based onthe proxies for expertise and tax avoidance used in this study. Finally, the findings documented in this study do not imply that industry knowledge is not important for auditors and tax advisors.
Section II briefly discusses prior research. Section III explains potential sources of bias in cross-sectional models of tax avoidance and auditor industry specialization. Section IV describes our methodology.Section V describes our sample selection and provides descriptive statistics on our samples. Section VI provides our results. Section VII includes a number of additional and sensitivity analyses. Section VIII concludes.
II. PRIOR LITERATURE ON TAX AVOIDANCE, AUDITOR PROVIDED TAX SERVICES AND AUDITOR INDUSTRY SPECIALIZATION
Researchers and the public alike have a keen interest in corporate tax avoidance, specifically, the amount of explicit taxes a company pays (Hanlon and Heitzman 2010). Corporate tax avoidance raises questions about companies paying “a fair share” of tax revenues, managers providing value to shareholders through tax savings, and, more generally, the various incentives and factors contributing to the cross-sectional variation in corporate taxes paid. In examining cross-sectional variation in tax avoidance, researchers commonly focus on variablessuch as size, return on assets, leverage, and foreign activities. Typically, these studies rely on book effective tax rates, cash effective tax rates, book-tax differences, and the residuals of a tax rate model as proxies for a client’s level of tax avoidance (e.g., Dyreng et al. 2008; Frank et al. 2009; Wilson 2009; Chen et al. 2010; Dyreng et al. 2010; andMcGuire et al. 2012)
Related streams of research focus on the effect of APTS on tax restatements (Kinney et al. 2004; Seetharaman et al. 2011), the impact of APTS on audit efficiency (Knechel et al. 2009), andthe relation between APTS and debt pricing (Fortin and Pittman 2008). The literature has also examined the relation between APTS and tax outcomes,such as tax accrual management (Cook et al. 2008), tax reserves (Gleason and Mills 2011), and more recently, tax avoidance (McGuire et al. 2012). Generally APTS studies find benefitsconsistent with potential knowledge spillover; however, Cook et al. (2008) and Knechel et al. (2009) find that APTS is associated with higher earnings management and less audit efficiency, respectively.
Extant research in auditing focuses on the role of the auditor as an expert or specialist. Evidence suggests that industry specialists,at the national and city-level, receive an audit premium for their services (Ferguson et al. 2003; Francis et al. 2005) and that audit quality is higher when an audit is performed by industry specialists (Balsam et al. 2003; Reichelt and Wang 2010). Recent studies also examine the interaction between APTS and industry specialization. Donohoe and Knechel (2009) find evidence consistent with an audit premium paid when industry specialists provide joint audit and tax services. McGuire et al. (2012) present evidence consistent with higher levels of tax avoidance in clients that purchase APTS from tax industry specialists compared to clients that purchase APTS, but do not use such specialists.
In contrast, we examine the relation between industry specialization and tax avoidance including non-APTS clients in our analyses. As discussed in the following sections, our analysis attempts to control for the influence of client characteristics and unobserved tax planning fees, to strengthen the inferences drawn from this line of research. Consistent with the prior literature, we rely on ETR, CETR, BTD and DTAX as proxies for tax avoidance. Moreover, we extend our analysis of APTS and industry specialization to examine tax aggressiveness. The literature has yet to link auditor specialization to more aggressive forms of tax avoidance. We use the level of unrecognized tax benefits (UTBSC) as a measure of tax aggressiveness, consistent with Lisowsky et al. (2012).
III. POTENTIAL SOURCES OF BIAS IN CROSS-SECTIONAL MODELS OF TAX AVOIDANCE AND AUDITOR INDUSTRY SPECIALIZATION
There are two main reasons as to why controlling for confounding factors is particularly important forstudying the effects ofauditor industry specialization. First, an audit firm may have extensive industry expertise even when its within-industry market share is small relative to other audit firms. Industry knowledge could be gained through other means; for instance, by the number of years an audit team has audited clients in the industry, by providing training to individual auditors, by auditing private clients in the same industry, by providing consulting services, or by hiring experts from within the industry or from other audit firms. Thus, it is not obvious that auditors with larger market share will have higher quality. Second, the evidence in Boone et al. (2010) and Lawrence et al. (2011) shows that the previously documented association between auditor size and audit quality could be attributed to differences in client characteristics, particularly to differences in client size. The separation of specialist and non-specialist auditors by within-industry market share also creates two groups of auditors with different client characteristics. For example, specialist auditors have larger clients compared to non-specialist auditors.In Table 2, Panel B we show that the mean size of the clients of specialist auditors (SIZE = 6.822 Column I) is 1.24 times the mean size of the clients of non-specialist auditors (SIZE = 5.487 Column IV).
Cross-sectional regression modelswith linear controls may result in inappropriate inferences dueto two problems that affect the research design simultaneously, selection bias and model misspecification. To overcome the selection bias, some studies (e.g., McGuire et al. 2012) use econometric designs that explicitly model the mechanism that results on differences in client characteristics between auditors, such as the Heckman (1979) self-selection model or two-stage instrumental variable models. A limitation of these research designs is that they require identifying appropriate exogenous instrumental variables or exclusion restrictions in the first stage, which is a difficult condition to meet in models predicting auditor choice (Lennox et al. 2012).[4] The variables that predict the choice of an audit specialist are strongly associated with all tax avoidance proxies. Matching constitutes an alternative to determine the auditor treatment effects, balancing the effect of client characteristics between specialist and non-specialist auditors.[5]