Answers are grounded in this paper and cite their sources.
Krishnan, Visvanathan, and Yu (2013) (hereafter, KVY) fill a relevant gap in the burgeoning auditor-provided tax services (ATS) literature by examining investor perception of the costs and benefits of ATS. Prior and contemporaneous studies have investigated the influence of ATS on tax avoidance, financial statement quality, and creditor perception, but a study of the influence of ATS on value relevance was missing from the literature until now. KVY provide evidence that the value relevance of earnings is positively associated with the ratio of tax fees to total fees paid to auditors.I begin my discussion at the very beginning of the paper; its original title was “Do Auditor-Provided Tax Services Impair the Value Relevance of Earnings?” When I first read KVY, the authors' use of the word “impair” in the title led me to believe that they hypothesize a detrimental effect of ATS on value relevance. However, the vast literature examining the influence of auditor-provided nonaudit services (NAS) on auditor independence and, by extension, financial statement quality (for an excellent review, see Schneider et al. [2006]) presents arguments for why ATS may either enhance or impair value relevance. On one hand, the joint provision of audit and tax services may facilitate knowledge spillover, whereby public accounting firms gain information about clients through tax consulting work that improves the quality of financial statement audits. On the other hand, paying tax fees to auditors may engender economic bonding (i.e., these payments compel auditors to acquiesce to clients' aggressive financial reporting and issue unqualified audit opinions). Given that both knowledge spillover and economic bonding arguments exist in the literature, I recommended that the authors modify the title to align with their hypothesis: “Do Auditor-Provided Tax Services Enhance or Impair the Value Relevance of Earnings?”While on the topic of economic bonding, KVY would benefit from expanded coverage of this issue, discussing the threat to auditor independence posed by economic bonding in a post-Sarbanes-Oxley (SOX) world. On one hand, SOX prohibits firms from purchasing most NAS from their clients. Simultaneously, audit fees have increased, on average, following SOX, and tax fees have declined markedly as a percentage of audit fees (Maydew and Shackelford 2007). Accordingly, NAS fees no longer provide the same magnitude of audit firm revenue that they did pre-SOX. On the other hand, most firms continue to purchase tax services from their auditors post-SOX, and tax fees are the largest (and, for most firms, the only) NAS to persist. Accordingly, tax fees may still constitute an important revenue stream for audit firms.I agree with the assertion in KVY that ATS are relevant to investors and boards of directors, particularly audit committees. I also believe that ATS are relevant to regulators and recommend that KVY provide some examples to support this contention. First, in 2008, Lynn Turner made a statement to the Advisory Committee on the Auditing Profession in which he recommended that the committee consider the “audit-only firm” as the appropriate business model for the audit industry (Turner 2008). More recently, in November 2011, the Internal Market Commissioner of the European Union proposed a law requiring Big 4 audit firms to separate audits from all NAS, including tax, to avoid conflicts of interest (CFO World 2011). Adding such examples would enhance the authors' contribution by demonstrating that ATS are still relevant to regulators subsequent to SOX and the follow-on regulations by the Securities and Exchange Commission (SEC) and the Public Company Accounting Oversight Board (PCAOB).After comparing the results of several studies probing the influence of NAS on investor perceptions, I question why various components of NAS influence earnings response coefficients (ERCs), market value of equity (MVE), and/or the value relevance of earnings differently. Specifically, Krishnan et al. (2005) find that the nonaudit fee ratio and the level of nonaudit fees are negatively associated with ERCs in 2001. In contrast, Lai and Krishnan (2009) find that MVE is greater for firms that purchased financial information systems services (FIS) from their auditors, and KVY demonstrate that ATS enhance value relevance. How should readers reconcile these results? Are the different conclusions attributable to different time periods, different response variables, components of NAS other than FIS and ATS, or some combination of these effects? One way for the authors (or other researchers) to address this issue would be to examine pre-2003 data. That is, data from the period before SOX banned most NAS would allow researchers to investigate whether and how various types of NAS influence value relevance and the relative magnitudes of these influences.Implicit in KVY is the assumption that investors are homogeneous. Stated differently, the authors do not examine cross-sectional variation in their results for different types of investors. But this variation may exist and shed light on the underlying drivers of the positive influence of ATS on value relevance. For example, PCAOB Rule 3522 asserts that auditors impair their independence by marketing aggressive tax positions to clients (PCAOB 2004). According, ATS may result in less aggressive tax positions than those suggested by nonaudit tax service providers, and transient investors may be more tolerant of aggressiveness than long-term investors. Similarly, sophisticated investors may be better able to assess the knowledge spillover benefits and economic bonding costs associated with ATS than naïve investors.To control for the various factors that influence firms' decisions to purchase tax services from their auditors, KVY use a two-stage model. In the first stage, the authors regress a dummy variable coded 1 for the existence of ATS on a set of 21 explanatory variables. I understand why most of these variables are associated with the magnitude of tax fees in general, but why are they associated with purchasing tax services from the auditor rather than some other provider? More discussion of why KVY include each of these variables would be helpful. Also, ten of these 21 explanatory variables fail to load significantly. Perhaps multicollinearity influences the results from estimating Model (1), or size usurps predictive power from other variables in the model. Either way, a more parsimonious model may be more appropriate.I question whether the positive association that KVY document between ATS and the value relevance of earnings is simply a function of earnings management. Krishnan and Visvanathan (2011) find that ATS are associated with less earnings management, and numerous studies document that earnings management reduces the value relevance of earnings in that managed earnings are valued less than unmanaged earnings (e.g., Gleason and Mills 2008). Accordingly, earnings management may mediate the relation between ATS and value relevance.The Securities and Exchange Commission (SEC) began requiring public companies to disclose ATS in 2003, and mandated disclosure of these fees for 2002 and 2003. In working on an ATS project with Tom Omer, we discovered that in many instances, Audit Analytics did not use firms' 2003 tax fee disclosures to backfill 2002 data. In the context of our project, we examined proxy statements for 555 firms for which Audit Analytics reported tax fees of $0 in 2002, but positive tax fees in 2003. Of those 555 firms, 532 disclosed positive tax fees for 2002 in their 2003 proxy statements. I recognize that KVY remove pre-2003 observations in a sensitivity analysis; however, I contend that because 2002 data from Audit Analytics are unreliable in the circumstance described previously, the authors should eliminate pre-2003 observations from their primary analyses.KVY identify a sample of firms that switched from their auditors to some other tax service provider, and find that value relevance declines from the year prior to the switch to the year of the switch. This supplemental analysis provides compelling evidence in support of the authors' main result. I encourage KVY to expand this change analysis in three ways. First, examine firms that hire their auditors as tax service providers to learn if such “reverse-switch firms” experience a higher value relevance of earnings. That is, does the documented switch effect work in reverse? Second, identify a sample of switch firms that subsequently switch back, and investigate if the magnitude of the decline in value relevance that the authors have already documented influences firms' decisions to rehire their auditors as tax service providers. Third, in the post-SOX era, if a firm uses its auditor as tax service provider and wishes to decouple these functions in order to preserve the appearance of auditor independence, the firm has two choices: (1) retain its auditor, but dismiss that audit firm as a tax service provider, or (2) dismiss its auditor, but retain that audit firm as tax service provider. Does the value relevance of earnings differ between these two groups? If knowledge spillover enhances audit quality and, in turn, financial statement quality, firms retaining their auditors may experience more value-relevant earnings than firms dismissing their auditors.As reported in Table 2, the mean of TAXFEE is more than ten times greater than the median. Because the distribution of TAXFEE is heavily skewed, outlying observations may influence the KVY results. Accordingly, I encourage the authors to truncate or winsorize their sample to guard against this possibility.Also in Table 2, the means and medians of Model (2) variables differ considerably between the ATS and non-ATS subsamples. Accordingly, perhaps KVY should include these variables in Model (1) as determinants of the decision to use the auditor as tax service provider.KVY fill a gap in the literature by examining the influence of ATS on the value relevance of earnings. However, the larger issues of (1) which types of NAS (audit-related, benefits, information systems, tax, and/or the residual “other”) affect the value relevance of earnings, (2) whether these effects are positive or negative, and (3) the relative magnitudes of these effects remain unanswered. The recommendations and concerns raised in this discussion indicate that future research is warranted to advance our knowledge of the influence of NAS on auditor independence and financial statement quality.
No takes yet. Share an insight, caveat, or question.
Kirsten A. Cook (2013) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: